GSM and Personal Communications Handbook
For a complete listing of the Artech House Mobile Communications Library, turn to the back of this book.
GSM and Personal Communications Handbook Siegmund M. Redl Matthias K.Weber Malcolm W. Oliphant
Artech House Boston • London
Library of Congress Cataloging-in-Publication Data Redl, Siegmund M. GSM and personal communications handbook / Siegmund Redl, Matthias Weber, Malcolm Oliphant p. cm. — (Artech House mobile communications library) Includes bibliographical references and index. ISBN 0-89006-957-3 (alk. paper) 1. Global system for mobile communications. 2. Personal communication service systems. I. Weber, Matthias K. II. Oliphant, Malcolm W. III. Title. IV. Series TK5103.483.R44 1998 621.3845'6—dc21 98-4710 CIP British Library Cataloguing in Publication Data Redl, Siegmund M. GSM and personal communications handbook—(Artech House mobile communications library) 1. Global system for mobile communications I. Title II. Weber, Matthias K. III. Oliphant, Malcolm W. 621.3’8456 ISBN
0-89006-957-3
Cover and text design by Darrell Judd. © 1998 ARTECH HOUSE, INC. 685 Canton Street Norwood, MA 02062 All rights reserved. Printed and bound in the United States of America. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the publisher. All terms mentioned in this book that are known to be trademarks or service marks have been appropriately capitalized. Artech House cannot attest to the accuracy of this information. Use of a term in this book should not be regarded as affecting the validity of any trademark or service mark. International Standard Book Number: 0-89006-957-3 Library of Congress Catalog Card Number: 98-4710 10 9 8 7 6 5 4 3 2 1
Contents
Contents
Preface
xv
Acknowledgments
Part I
1
xxi
GSM in the light of today
1
The changing scene—again
3
1.1
The digital cellular evolution
4
1.2
Basic market figures and the system standards
6
1.2.1 Cellular and personal communications services: market presence and potential
10
1.2.2
13
1.3
Meeting the demands
Aspects on marketing the product
17
1.3.1
Service providers
18
1.3.2
Fulfillment houses
20
1.4 Phones: shrink them, drop their price, and grow their features 1.4.1
What’s your size?
20 21
v
vi
GSM and Personal Communications Handbook
1.4.2
How long can you stand by?
21
1.4.3
Ninety-nine cents?
22
1.4.4
What can you do that I can’t?
23
1.4.5
Multiple bands and multiple modes
23
What is personal communications?
26
1.5
1.5.1
PCS: defining the requirements
27
1.5.2
PCS: the technical solutions to the requirements
30
1.5.3
PCS and what system technology?
36
1.5.4
Where does it lead?
37
1.5.5
GSM and PCS in the United States: an overview
42
References
2
x
From Pan-European mobile telephone to global system for mobile communications 51 2.1 GSM: what it was meant to be and what it became
52
2.1.1
The initial goals of GSM
52
2.1.2
The initial results
52
2.1.3
First experiences
54
2.1.4
PCN networks and DCS 1800
55
2.1.5
PCS 1900
59
2.1.6
UIC
63
2.2
The role of the GSM MoU
65
2.3
ETSI and the Special Mobile Group
67
2.4
Standards: the present and the future
69
2.4.1
GSM Phase 1
72
2.4.2
GSM Phase 2
72
2.4.3
GSM Phase 2+
74
2.5
GSM type approval issues
75
2.5.1
The objectives
77
2.5.2
The authorities
78
Contents
vii
References
3
A look over the fence 3.1
Competition or complement?
81 83
3.1.1
Cellular and personal communications
83
3.1.2
Cordless access
84
3.1.3
Wireless in the local loop
85
3.2
What else is out there?
86
3.2.1
Digital Enhanced Cordless Telecommunications
88
3.2.2
Personal Handy Phone System
96
3.2.3
Personal Access Communications System
96
3.2.4
CDMA (IS-95)
101
3.2.5
TDMA (IS-136)
104
3.2.6
IS-661
111
3.3
Noncellular digital trunking systems
117
3.4
Interference and health issues
122
References
Part II
4
x
GSM services and features The development of GSM standards and features
4.1
Phase 1
125
127 129 132
4.1.1
Phase 1 teleservices
132
4.1.2
Phase 1 bearer services
132
4.1.3
Phase 1 supplementary services
133
4.2
Phase 2
134
4.2.1
Phase 2 teleservices
134
4.2.2
Phase 2 supplementary services
135
4.2.3
Phase 2 network improvements
136
viii
GSM and Personal Communications Handbook
4.3
Phase 2+
4.3.1
Release 96
138
4.3.2
Release 97
143
4.4
Conclusion
References
5
138
GSM telecommunication services
144 146
147
5.1
Bearer services in GSM
149
5.2
Teleservices in GSM
152
5.3
Connection types in a GSM PLMN
152
5.3.1
Lower layer capabilities
153
5.3.2
Connections
153
5.3.3
Attributes between two networks
155
5.4
Rate adaptation
157
5.4.1
Error protection
163
5.4.2
Terminal equipment and mobile termination
164
5.5
Radio link protocol
165
5.5.1
Frame structure
166
5.5.2
Control of RLP
167
5.5.3
Error recovery
171
5.5.4
RLP summary
171
5.6
Access to different networks
172
5.6.1
Transmission into the PSTN
172
5.6.2
Facsimile transmission
174
5.6.3
Transmission into the ISDN
175
5.6.4
Transmission into the PSPDN
175
5.6.5
Transmission into the CSPDN
177
5.7
Fax services
5.7.1
End-to-end view via the GSM infrastructure
178 179
Contents
ix
5.7.2
Configuration at the mobile station
181
5.7.3
Transparent fax service
182
5.7.4
Nontransparent fax service
185
5.7.5
In-call modification
186
5.8
Connecting a mobile station to external devices
5.8.1
Application for short message services
188
5.8.2
Remote control of mobile equipment
190
5.9
Future developments
193
5.9.1
High-speed circuit-switched data
194
5.9.2
General packet radio service
199
5.9.3
Packet data on signaling channels
202
5.9.4
The 14.4-Kbps user data rate
204
5.9.5
Facsimile enhancements
204
5.9.6
General bearer services
205
5.9.7
Emergency call with additional data transfer
206
References
6
187
Short message service 6.1
Short message service: point to point
206
211 212
6.1.1
Implementation of point-to-point SMS in the network
213
6.1.2
Alphabet of SMS
228
6.1.3
Example of a SMS-MT message frame
228
6.1.4
Problems that can occur while sending short messages
231
6.1.5
SMS and supplementary services
232
6.1.6
Use of additional devices for SMS
233
6.1.7
The future
235
6.2
SMS cell broadcast
237
6.2.1
Implementation of CB in the network
238
6.2.2
Contents of a cell broadcast message
240
6.2.3
Future developments for cell broadcast
243
References
244
x
GSM and Personal Communications Handbook
7
Supplementary services 7.1
Introduction to supplementary services
245 246
7.1.1
Network entities
248
7.1.2
Password handling
251
7.2
Call forwarding supplementary service
252
7.2.1
General behavior of call forwarding services
253
7.2.2
Operation of call forwarding
255
7.2.3
Conflicts for call forwarding
259
7.2.4
Who pays for what?
260
7.3
Call barring supplementary services
262
7.3.1
Call barring for incoming and outgoing calls
263
7.3.2
Applicability of call barring
264
7.3.3
Restrictions to call barring
265
7.4
Line identification supplementary services
266
7.4.1
Calling line identification
266
7.4.2
Connected line identification
268
7.5
Call waiting
268
7.6
Call holding
271
7.7
Multiparty communication supplementary service
272
7.8
Advice of charge supplementary service
275
7.8.1
Charge advice information
276
7.8.2
Advice of charge (information)
277
7.8.3
Advice of charge (charging)
278
7.9
Closed user group supplementary services
279
7.10
Unstructured supplementary services data
281
7.11
Implementation of SS in a GSM mobile station
283
7.11.1
Implementation of non-call-related SS
284
7.11.2
Implementation of call-related SS
288
Contents
xi
7.11.3
Implementation into a menu structure of an MS
288
7.12
Additional implementations in the mobile phone
289
7.13
Future developments for Phase 2+
290
7.13.1
Call deflection
291
7.13.2
Call forwarding enhancements
291
7.13.3
Call transfer
291
7.13.4
Call completion services
292
7.13.5 Direct subscriber access and direct subscriber access restriction
295
7.13.6
Malicious call identification
295
7.13.7
Mobile access hunting
296
7.13.8
Support of private numbering plan
296
7.13.9
Multiple subscriber profile
296
7.13.10
Universal access to freephone numbers
297
7.13.11
Premium rate service
297
7.13.12
Charging
298
7.13.13
User-to-user signaling
299
References
8
The subscriber identity module
300
303
8.1
Memory structure
305
8.2
Security
306
8.3
Phase 1 SIM
309
8.4
Phase 2 SIM
310
8.5
Phase 2+ SIM
323
8.6
The SIM initialization process
332
8.7
Electrical characteristics of the SIM
333
8.7.1
SIM Power Supply
333
8.7.2
SIM memory
334
8.7.3
SIM architecture
336
8.8
Outlook for future applications
338
xii
GSM and Personal Communications Handbook
8.8.1
NATELsicap by Swisscom
339
8.8.2
Prepaid SIM
340
8.8.3
Future parameters
343
References
9
New Phase 2+ functions 9.1
SIM application toolkit
345 346
9.1.1
Overview of the SIM application toolkit
346
9.1.2
Profile download
347
9.1.3
Proactive SIM
347
9.1.4
Data download to SIM
350
9.1.5
Applications using the SIM application toolkit
353
9.1.6
Conclusion
357
9.2 Customized applications for mobile network enhanced logic (CAMEL)
357
9.2.1
Functional description of CAMEL
358
9.2.2
Network architecture
359
9.2.3
A CAMEL example
360
Railway applications
361
9.3
10
344
9.3.1
Enhanced multilevel precedence and preemption
361
9.3.2
Voice group call service
365
9.3.3
Voice broadcast service
368
Refernces
369
Roaming and call routing
371
10.1
Routing in GSM PLMNs
372
10.1.1
Location registration
372
10.1.2
Routing within a PLMN
375
10.1.3
Call routing when a mobile station is roaming
376
10.2
Charging principles
378
Contents
xiii
10.2.1
National call charges
378
10.2.2
Call charges when roaming
379
10.2.3
Call forwarding
380
10.2.4
More exceptions to the rule
380
10.3
Phase 2+: support of optimal routing (SOR)
10.3.1
Roaming mobile subscriber
382
10.3.2
Call forwarding to home country
382
10.3.3
Call forwarding to visited country
384
10.4
Conclusion
References
Part III
11
381
GSM technology and implementation Introduction to GSM technology and implementation
11.1
Breaking GSM down
384 385
387 389 391
11.1.1
Physical and logical blocks of a GSM mobile station
391
11.1.2
Physical and logical blocks of a GSM base station
396
11.2
Transmitters and receivers
397
11.2.1
Transmitters
398
11.2.2
Receivers
402
11.3
MS and BTS—new roads to the ultimate radio
410
11.4
Baseband signal processing
412
11.5
Speech coding and speech quality in GSM
415
11.5.1
Speech coding tutorial
415
11.5.2
Speech quality
422
11.5.3
DTMF and signaling tones
423
11.5.4
GSM full-rate speech coding
424
11.5.5
GSM half-rate speech coding
424
11.5.6
GSM enhanced full-rate speech coding
425
xiv
GSM and Personal Communications Handbook
11.5.7
Complexity comparison FR-HR-EFR
427
11.5.8
The future for GSM speech coding
427
11.5.9
Speech coding and …
429
11.6
Equalizers
433
11.6.1
The problem—ISI
435
11.6.2
General equalizers
441
11.6.3
Viterbi equalizer
444
11.7
Encryption and security in GSM
459
11.7.1
Algorithms and keys
459
11.7.2
Ciphering in GSM
460
11.7.3
Regulations
461
11.7.4
Security vs. fraud
461
11.8
Mixed signals
462
11.9
Microprocessor control
465
11.10
GSM timing
466
11.11
Components and technology
468
11.12
Guide to the literature
470
11.12.1
General radio design
470
11.12.2
Coding and its mathematics
470
11.12.3
Digital radio
471
References
471
Appendix: Coding of the default GSM alphabet
475
Glossary
477
About the authors
499
Index
501
Preface
Preface
U
se of the global system for mobile communications (GSM) continues to spread throughout the world. It works, it is efficient, and it is well liked. As is true of any mature but vital and growing system, the services and equipment based on the GSM specifications are still evolving to accommodate its new users and operating environments. The new services, improvements, applications, and products are the new flavors offered in the GSM ice cream stand of wireless telecommunications networks. New terminals feature increased standby and talk times while their sizes shrink and their prices fall. The combination of competitive pricing and access to a growing menu of services, which is attractive to a wider variety of users, marks the transition of GSM from a high-end offering to a consumer product orientation. Interesting features that go far beyond the point-to-point voice conversation link typical of traditional wireless services are a reality in GSM networks. Sophisticated data services with access to the Internet, video connections, ISDN links, and supplementary services, which are expected in wireline digital networks, are becoming a reality in GSM-based networks. Why have the authors of the previous work An Introduction to GSM (Artech House, 1995) chosen to write again on the same subject? Considering the metamorphosis just described, one answer is clear: GSM remains an evolving standard. Its use and applications are no longer
xv
xvi
GSM and Personal Communications Handbook
restricted to the initial aims of the early experimenters, the mothers and fathers of GSM. Another reason appears when we compare the coverage of our first book with this volume. An Introduction to GSM was meant to be—and was widely accepted as—a first confrontation with the subject. It focused on what GSM is rather than on how it works and how it can bring a growing catalog of services and applications to its users. The subject was treated so as to explain the system architecture and radio techniques used to convey information from one point to another. Some chapters on testing—still an important subject, especially because GSM is supposed to be an open standard—were added to enhance understanding, and further illustrate the techniques and processes. The first book, then, can be considered an illustrative brochure describing GSM to those considering its purchase and use. This volume is a logical extension of the earlier book; it is a user’s manual for those who wish to exact efficiency and new features from GSM. When An Introduction to GSM came to life, GSM services, as well as those of its digital cellular system (DCS) cousin, were struggling in their infancy. Networks were just starting to offer their services based on the status of the standards termed, at that time, GSM Phase 1. The situation has completely changed today. GSM has moved from the showroom to the customer’s garage. Accessories have been added and copies have been made. Today we see three different standards based on the core GSM technology: GSM 900, DCS 1800 (GSM 1800), and PCS 1900 (now called GSM-NA). Today, network operators are introducing GSM Phase II services and products. Noting that GSM has found favor far beyond its original European roots, researchers and developers in the industry and standardization bodies are working on GSM Phase 2+ services, features, and products. What are these new services, what do they offer, and how do they work? The world has adopted GSM as the most widely deployed digital cellular standard with expanding interworking and roaming capabilities. With wider deployment comes greater variety. Since the publication of our first book, we have seen other systems employing different wireless access techniques adopt key GSM properties. The number of new applications, varieties, and flavors of GSM grows with its acceptance. As the world’s dominant wireless protocol, the industry has accepted the responsibility of exploring ways in which GSM can interwork with other wireless access systems. Success in these efforts will yield a platform
Preface
xvii
capable of even greater capacity for the improved services and bandwidth allotted to carry them. Still, GSM is not the only technology setting out to win the hearts and cash of those who want to use or offer digital wireless communications services. Competitive technologies such as code division multiple access (CDMA) are poised to take their share of the market as they offer their own set of applications. Whatever these systems may be—cellular radio, personal communication systems (PCS), specialized mobile radio (SMR), wireless in the local loop (WLL), cordless phones, or even satellite-based systems—GSM will thrive. The better these systems work together for the benefit of their users, the more all the providers, whatever technology they select, will win. What are the important issues that drive the variations and added features? What are the trade-offs and compromises? What are the limitations? Where are the solutions? These matters are treated in the widely accepted style of the authors’ first GSM volume. The answers are in the details. Subjects like digital baseband technology, new radio techniques and implementation schemes, and intelligent networks are shrouded in specialized language and mathematics. This increased specialization, which is not unique to mobile radio, frustrates managers, marketing specialists, and others taxed with the responsibility of financing and deploying GSM networks, and building the devices and the equipment on which these networks depend. Technical specialization tends to stifle effective communication among people. Just as with our first book, this one is written for those who must manage GSM projects and the growing variety of technical specialists needed to run them. Jargon and specialized mathematics are avoided, and new terms are explained as they are introduced. Your authors have worked diligently to explain obscure but important concepts, processes, and devices in clear language without the aid of sophisticated mathematics shorthand. Moreover, the explanations are animated with some of the excitement and passion of the scientists and engineers. The actual work must, however, be left to the technologists and the specialists, for it is only with the appropriate tools (mathematics, software, jargon, and experience) that the devices and features in the GSM networks can be manipulated efficiently and designed at a price such that many people can afford them. The treatments in this book, therefore, include appropriate references for each of the subjects covered. Some of the matters, particularly the radio techniques covered in the last
xviii
GSM and Personal Communications Handbook
chapter, are so vast in their scope that small tutorials to carefully selected references are included. Just like our first book, this one is an initial confrontation, a guide, or an orientation for further reading. Students and engineers new to GSM and digital mobile radio will, therefore, find this book helpful. Even as we finished the chapters, we knew there was still lots to tell about GSM. This book illuminates just another episode in its continuing story. The reader is invited to view this latest volume as a complement to An Introduction to GSM. Though some of the chapters make direct reference to the earlier book, anyone familiar with the fundamentals of GSM will be rewarded here. This book has 11 chapters sorted into three parts. Part I has three chapters. The first chapter is a status report on where GSM is deployed in the world today and where it is likely to be accepted in the future. Some market sizes and other figures are offered. The confusion over the PCS and PCN designations is explored in the light of competing wireless systems. Chapter 2 traces the phased deployment of GSM. A means toward understanding current and future enhancements and system variations is possible when we understand how GSM adapted to conflicting national and regional requirements from its original, narrow European goals. Chapter 3 looks at the influence of new technologies, such as CDMA, other wireless services, such as TETRA, and certain social and economic realities, such as the North American market, on GSM as well as how GSM tempers those influences. Part II consists of Chapters 4 through 10, which explain in detail the huge number of features and services of GSM. GSM introduces user services and improvements in phases. Chapter 4 starts by describing these services as it traces their phased introduction into the networks. Chapter 5 describes how teleservices and bearer services are handled in networks. Chapter 6 covers short message service (SMS), a service that is not found in wireline networks and a popular feature that removes so much of the intrusive nature of basic cellular service. Chapter 7 explores supplementary services (SS), which are those features common in many digital wireline networks that users find so helpful in their busy lives. Caller ID is an example. The subscriber identity module, also referred to as SIM card, which is viewed with growing interest and envy by proponents of some newer competing mobile radio technologies, is thoroughly explained in Chapter 8. Chapter 9 covers the latest features introduced into GSM, such
Preface
xix
as the SIM application toolkit, CAMEL, and features introduced for railway applications. Chapter 10 brings us back to the network side, describing parameters stored in various switches or registers. This discussion reveals the mysteries of when and how the charging clock ticks and how calls are actually routed through the network. Part III concludes the book with only one chapter, Chapter 11, which covers some salient technology issues. The emphasis is on the handsets. GSM handsets are the most visible part of the network, and their variety and high quality have played a major role in the acceptance of GSM in so many markets. The handset’s ability to carry all the features described in Part II is seen as a miracle by many of the people who purchase these wireless wonders in all their colors and shapes.
Acknowledgments
T
he authors wish to acknowledge the help of Dr. William R. Gardner of LSI Logic whose advice on the details in Chapter 11 was critical to our work. We also thank Vinay Patel of Hughes Network Systems and Mark Vonarx of Omnipoint for their generous help with Chapter 3. The authors also acknowledge the enduring patience and support of their employers. A sincere thank you also goes to the personnel at Artech House for their help in bringing this book into existence. Many thanks go to the unknown reviewer(s) for their excellent work and many important inputs, which improved the quality of all the chapters. Siegmund Redl offers his special thanks to his wife Johanna and son Christoph for their patience and support. Matthias Weber likewise extends special thanks to his wife Ilse and daughter Laura, who endured his sporadic hours, days, and weeks of absences while completing this work.
xxi
PART
I GSM in the light of today
CHAPTER
1 Contents 1.1 The digital cellular evolution
The changing scene—again
The changing scene—again
1.2 Basic market figures and the system standards 1.3 Aspects on marketing the product 1.4 Phones: shrink them, drop their price, and grow their features 1.5 What is personal communications?
T
o the delight of its supporters and the surprise of its detractors, the global system for mobile communications (GSM) has, after a few false starts and sputters, found its place in the communications world—and what a place it is. GSM has brought low-cost and reliable mobile communications to most of the countries of the world. Its features and options are rich enough to satisfy the peculiar and disparate needs of the users in all of the GSM countries. A system definition that came to life through a Pan-European initiative for a single open cellular standard has spilled over the borders of Europe and continues to conquer new territory around the world. Not only is new territory conquered geographically, but also includes new user groups within countries (target customers), new services, and new applications. What happened? Where did all these 50 million phones and users come from only five years after the introduction of the first GSM-based cellular services?
3
4
GSM and Personal Communications Handbook
What were the initial intentions for establishing the standard, and what were the early experiences? How did the GSM standard succeed in Europe only to then spread into new areas, new frequency bands, and new applications, and how will GSM continue its march around the globe? What is GSM’s future? What is GSM’s place in personal communications systems (PCSs)? And what is a PCS anyway? In this and in subsequent chapters, we will find answers to these questions. GSM is not what it used to be only five years ago, and it will change into something else next year. In this chapter we pause to see what GSM means today. We review market figures and perspectives, and discuss new applications and marketing schemes. We also dig into some technology and standardization issues, and consider key definitions of services and features around GSM.
1.1 The digital cellular evolution With the introduction of digital technology as the second generation of wireless communications, the world experienced a tremendous growth in cellular subscribers. The potent combination of added value service features and applications, increased capacity, expanded coverage, and improved quality of service continues to make mobile telephony an indispensable commodity. Once users try reliable private communications unconfined by wires, they do not want to give it up. Deregulation, growing demand, marketing prowess, competition, and open standards are the pillars on which the success of digital cellular rests [1]. Today, the term personal communications usually applies to new systems and services (PCSs) that are offered to an increasing portion of our world’s population. The underlying technologies for personal communications are digital cellular ones, with GSM being the most widely accepted form today and for many years to come. But what is the difference between PCS and cellular? One way to clear the confusion is to regard PCS as a deployment scheme that occurred after cellular. Because the deployment added significant additional wireless capacity, something had to attract new subscribers to fill
The changing scene—again
5
the additional capacity: if cellular was for the business users and the wealthy, then PCS was for the mass market. From a pure technology perspective, personal communications describes a set of services that a customer might expect. Cellular refers to a range of technological solutions that may be used to deliver such services. The vast majority of PCS subscribers are receiving these services by using a cellular technology. However, there are a minority of “noncellular” technologies that can also be a basis for personal communications services, for example, cordless systems. Some PCSs may use more than one technology to deliver a comprehensive service. For now, let’s consider PCS as an approach that can bring mobile communications services to a broader consumer market. PCS originally stood for a North American initiative with new spectrum allocations in the 1900-MHz band, thus the term PCS 1900 for the GSM 900 derivative in North America. The original PCN term was introduced in the United Kingdom before PCS, and referred to the personal communications networks licensed in the 1800-MHz band; whereas the PCS term originated in the United States and originally referred to spectrum licenses auctioned in the 1900-MHz band. As such, the United Kingdom hosted the world’s first PCSs, which were—and still are—referred to as PCNs, personal communications networks. One alternative for the awkward PCS 1900 term is GSM-NA, which stands for GSM North America. For the general user who neither knows nor cares what a hertz is, the new 1900-MHz allocations are simply extensions to the cellular network. But because the new frequency allocations were auctioned off to new operators, fierce competition in certain trading areas arose for new subscribers. Once we review the whole market, a general approach for defining personal communications services and related marketing efforts is discussed later in this chapter. Eventually, innovative marketing concepts shake each other out, buzzwords settle in our minds, and the imaginative blurs into the familiar. What finally counts in the end is what service we can get and what it will cost. Before we commence our discussion of the different flavors of personal communications through digital cellular or simply wireless access systems, we need to look at how the market developed its services and their prospects for the future. Then, we need to look at GSM’s dominant place in the picture.
6
GSM and Personal Communications Handbook
1.2 Basic market figures and the system standards The dynamic market for personal wireless communications shows dramatically increasing growth rates. At the end of 1995 more than 85 million users were subscribing to cellular telephone services. The year 1996 saw more than 133 million users. Predictions for the year 2000 and beyond are constantly continuing to rise, and they currently range between 300 and 500 million subscribers. For comparison, in 1996 the world’s population (more than 5 billion) had only about 800 million fixed-line telephone installations, which was a 16% penetration. The cellular penetration accounted for only around 2% of the world’s population. Cellular subscribers, worldwide, exceeded the 100 million mark in 1996 with plenty of momentum to achieve multiples of this figure by the turn of the century. Penetration in some countries (Scandinavia and Australia) is already at 30% with a trend toward 40%. Another trend is for revenues from mobile services to exceed those achieved by fixed-line services even though the amount of traffic generated through mobile phones is much less than in the fixed networks. Private mobile radio (PMR)—which is increasingly dominated by socalled trunking systems, cordless telephony, paging, and messaging— wireless local-area networks (WLANs), and wireless in the local loop (WLL) are other wireless business sectors that either saw a proportional growth with cellular or still have a huge growth potential. The growth potential is particularly bright for WLL, because many developing regions are now provided with flexible, uniquely tailored, and cost-effective wireless access to telephony services. Deregulation, new spectrum allocations, and new network operators will install WLL systems to provide services in developed countries too. The fixed-wire plant is a very expensive structure that requires lots of maintenance and is sensitive to storms and vandalism. The “last mile” of the system, which provides the access to the home and office, as well as installations within offices and factories, is often more efficiently covered by radio transmission. Whereas the majority of users still subscribe to analog cellular networks (67 million or 79% in 1995 [2]), digital systems are catching up. It is expected that in the year 1999 about 80% of the new mobile phones sold to customers will be based on some kind of digital technology.
The changing scene—again
7
Figure 1.1 depicts the growth of digital cellular subscribers—also at the expense of analog—from 1991 to 2000 (the figures for 1996 through 2000 were estimates) [2]. The analog systems comprise the wellestablished technologies: AMPS, TACS, NMT (450, 900), as well as some other minor systems. The digital systems include GSM and its derivatives (digital cellular system DCS 1800 or GSM 1800 and PCS 1900 or GSM-NA), dual- and single-mode digital AMPS or TDMA (in the United States this is called D-AMPS or Interim Standard IS-54/136), personal digital cellular (abbreviated PDC in Japan) and code division multiple access (CDMA; also called IS-95). AMPS-based systems with a digital overlay (TDMA or CDMA) are widely deployed in North America and in some South American countries (Argentina, Brazil, Peru, etc.) to take advantage of the large number of AMPS phones already in use in these regions. Some Asian countries and a few other regions employ the same scheme on a diminished scale. The CDMA version is found very successful in South Korea and Hong Kong, and the TDMA version in Israel. We can expect to find such hybrid systems wherever a successful AMPS system is already deployed. IS-95-based CDMA will be seen in Japan, as this technology was chosen to be overlaid with existing analog (TACS-based) technology in order to increase network capacity (see discussion in Section 1.2.1). GSM-based systems are not hybrid ones, and they are found almost everywhere: Europe, the Middle East, Africa, many Asian countries, and Australia. In some regions we find a mixture of cellular systems that 300 250 200
Total Analog Cellular Subscribers Worldwide (Millions)
150
Total Digital Cellular Subscribers Worldwide (Millions)
100 50 0 91
93
95
97
99
Figure 1.1 Subscribers in analog versus digital cellular networks worldwide (From: [2]).
8
GSM and Personal Communications Handbook
simply coexist next to each other. Australia has both AMPS and GSM systems, and the AMPS system has been decreed to be taken out of service by 2000. No match can be found for the cellular salad in Hong Kong, where in 1997 we found the following system standards in operation: AMPS, CDMA, D-AMPS/TDMA, GSM, TACS, and CT2 (telepoint). Additional PCS licenses shall be awarded. The Scandinavian countries (Sweden, Norway, Denmark, and Finland) have cellular penetration rates between 20% and 30%, and some are on their way to 40% in 1997. The Scandinavian trend is a predictor for the rest of the world. Worldwide, there will soon be more new subscribers signed up for services supplied by wireless systems than are being connected to pure wireline networks. Cellular service will replace certain fixed-line services for a variety of applications. GSM networks will prevail in the year 2000, with a conservative estimate of 157 million subscribers representing more than 43% of the total cellular market [2]. GSM will grow with the general market. Figure 1.2 displays the worldwide trend with GSM’s share in subscriber numbers. As was the case in Figure 1.1, the figures for 1996 through 2000 were estimates. The split between digital technologies (GSM, CDMA, IS-54/136, and PDC) predicted by one source [2] is typical of all those who analyze the market. The subscriber share will be about 57% for GSM, 22% for CDMA, 15% for TDMA (IS-54/136), and 7.5% for PDC [2].
400 350 300
Total Subscribers Worldwide (Millions)
250 200
GSM Subscribers Worldwide (Millions)
150 100 50 0 91
93
95
97
99
Figure 1.2 GSM subscribers versus total (analog and digital) cellular subscribers worldwide (From: [2]).
The changing scene—again
9
So, CDMA technology (U.S. Interim Standard IS-95, which is also applicable for North American PCS as SP-3384) is coming in second according to the current estimates of market penetration. Though there have been some stunning successes, particularly with PrimeCo (a large consortium of PCS operators in the United States), this technology is several years behind its digital rivals in terms of system deployment and product availability. Claims of CDMA’s superiority and its potential to become a world standard have been constantly reduced to more realistic views in the past 3 years. With every year of delay in the broad introduction of CDMA-based services, and with more understanding of the cold technical issues and problems, the industry backers had to eventually realize that CDMA was not the magic solution for all wireless markets. The initial wild enthusiasm was replaced with a sober consideration of CDMA’s real advantages, for example, a substantial reduction in frequency planning tedium and some new vocoder technology. As the industry turns its attention from exciting marketing promises to tedious engineering reality, it will discover how to take advantage of CDMA’s benefits and improvements over today’s TDMA technologies, and a great potential in many markets may be realized. When functional networks and attractive products finally become available at the right times, with adequate quality, and in the correct volumes, CDMA will have its day. Unlike the situation in South Korea where CDMA is the official cellular technology, and the European situation where GSM is the decreed protocol, North America was and still is a battlefield between GSM, CDMA, and IS-136. More than 50% of the new PCS operators in the United States (covering an adequate potential subscriber base) have decided to and have already started to deploy IS-95 CDMA-based technology. The remainder is GSM and IS-136 territory. The mathematics behind how many “pops” are covered by which network operator and therefore by which technology is sometimes confusing. Let us stick with the simple statement that, for North American PCS, CDMA takes the lead, followed by GSM-NA and then IS-136. The American regulators take no sides. The new PCS operators need to sort through all the proposed technologies for which sound arguments can be assessed. The religious has to eventually give way to the practical, claims have to be proved, and time has to be the arbiter. Decisions on which technology to support and deploy are also dependent on intellectual
10
GSM and Personal Communications Handbook
property rights (IPRs) and patents. GSM is typical of a system in which IPRs are shared among industry players who agree on certain conditions for licensing them. When agreement is reached among the participants, then everyone can participate in the system’s development. Even though IS-95 is an open standard, most of the IPRs for CDMA are in the hands of just a few companies and individuals. When most of the IPRs are owned by a single player, innovation tends to be stifled because licensing restrictions cramp the resources of players who could otherwise make important contributions to the development of a system. A balance needs to be found between the valid interests of IPR holders who want to collect licensing fees, and the industry that needs to design, build, and market low-cost mass products. Worldwide, more than 40 major telecommunications equipment manufacturers have licenses for IS-95 CDMA. The balance is tested in the achievable volumes of products. Volumes, however, can only be achieved with clear industry commitment, functional products sporting attractive services, competitive pricing, and an early market presence. Claims that the sheer use of a certain technology will make things work out well do not count when it comes to investing large amounts of cash. Because CDMA is an innovative system worthy of serious consideration, it will be refined and a balance in recovering the costs of its development will be struck. GSM also has its IPRs whose holders are spread throughout the industry. Cross-licensing of patent rights is common practice among the holders. Eventually, CDMA’s proponents and supporters may join forces to contrive a way in which it can coexist with the huge (and still growing) GSM deployments already in about 100 countries. If and when they do, CDMA will grow to be a major carrier of cellular traffic. 1.2.1 Cellular and personal communications services: market presence and potential
We just need to look at Figure 1.3 (the present) and Figure 1.4 (the future), taking into account subscribers on all cellular-based technologies including PCS and PCN, to see which regions have grown their subscriber bases. As you examine the figures, please note that there is a general growth of the whole cake in Figure 1.4 by a factor of approximately 4! Please also note that some regions are growing faster in population than others, and that the whole demographic behavior of such a wireless
The changing scene—again
11
1995 Cellular Subscribers Worldwide Asia and Australia
Japan
Central and Eastern Europe
Middle East and Africa
Latin America
Western Europe
North America Total is 85 million
Figure 1.3
Cellular subscribers worldwide in 1995 (From: [2]).
2000: Cellular Subscribers Worldwide—estimate Japan Middle East and Africa
Asia and Australia
Western Europe Central and Eastern Europe Latin America Total is 363 million North America
Figure 1.4 Cellular subscribers worldwide (including PCS) in 2000 (From: [2]).
subscriber base system is very dynamic in several planes, and difficult to model and describe in words. We want to look at two snapshot sample situations here, the present and the year 2000, with predictions of market evolution over the next 5 years.
12
GSM and Personal Communications Handbook
W e s t e r n E u r o p e a n d N o r t h A m e r i c a Today, Western Europe and North America account for the majority of cellular subscribers worldwide. With different well-accepted analog standards such as AMPS, NMT, TACS, Radiocomm 2000, and the C-Net system, and with the introduction of GSM in Europe with a strong foothold in countries such as the United Kingdom, Germany, Italy, and France, Western Europe had approximately 23 million subscribers in 1995. This was 27% of the worldwide subscriber base. The expected growth in Europe is predicted to be almost 80 million subscribers by the year 2000. As large as this number is, it accounts for only 22% of the increased worldwide subscriber base by 2000 [2]. This is due to the explosive growth in other regions. In 1995, the United States and Canada had 35 million subscribers, which accounted for more than 40% of all the world’s cellular users. Even as market penetration will grow substantially into 2000, the proportion of the world’s cellular users represented by North America will shrink.
Asia, especially the “four tigers” (South Korea, Singapore, Hong Kong, and Taiwan), and Japan come in third and fourth, respectively, in terms of market size in 1995, but dominate the scene in 2000. Japan, being a special case, accounted for enormous growth rates in the recent past and will fuel even more growth in the future. Since its deregulation in 1994 the Japanese market grew quickly from 8 million cellular subscribers in 1995 (already with 40% digital subscribers on personal digital cellular [PDC] systems) and will expand further to more than 25 million subscribers in 2000. There is a good potential for the future in Japan [2]. The traditional revenue per subscriber in Japan was, until recently, relatively high due to high tariffs, which often included leases of the terminal equipment. Deregulation and market liberalization in 1994, which allowed competition to Nippon Telephone and Telegraph (NTT DoCoMo) and the sale of handsets, sparked a tremendous run on wireless services. Besides the success of cellular (HCAP/J-TACS/N-TACS/N-MATS and PDC 800 and 1500), the Personal Handy Phone System (PHS), which provides two-way telepoint service, is chiefly responsible for the enormous success of personal communications in Japan. After the start of services in mid-1995, there was a steady growth in subscribers up to 3.5 million in mid-1996. Expectations for 1997 are as high as 8 million with a steady and sustained increase beyond the year 2000 [3]. This enormous acceptance by mobile users is due to good coverage (mainly in cities), low Asia and Japan
The changing scene—again
13
prices and cost of ownership, good service quality, and popular products. PHS competes very well with cellular in Japan. Cellular is doing very well in Japan too, so well that operators fear they will run into capacity problems. This is why they investigated the use of a CDMA overlay to the current J-TACS and PDC systems. Personal communication services are also deployed by four operators in Korea in the 1800-MHz band, based on IS-95 CDMA. The biggest potential in the rest of Asia is found in highly populated countries where fixed-line penetration is low and wireless access (through cellular and WLL) is a decisive economic factor that governs growth. E a s t e r n E u r o p e There is a lot of market potential in many Eastern European countries. Since the fall of the Iron Curtain, we have seen a plethora of regulating efforts and commercial enterprises spring up that target the supply of wireless access technology for telecommunications services. 1.2.2
Meeting the demands
We sense that people naturally tire of being restricted to a wired infrastructure—and that once they try mobile radio they will not give it up. This is not the whole story, for you cannot try something to see if you like it unless you have the means and reasons to do so in the first place. We need a more disciplined examination of what people do with mobile phones and how the services appear. To gain an understanding of the growing demand for wireless communications services, we have to look at the way access to communications services is distributed, and then note the demand for those services in the light of telecommunications marketing practice. We also need to distinguish between two separate cases in our investigations: (1) the economically developed countries and (2) the developing countries. 1.2.2.1
Developed and developing countries
In economically developed countries, mainly North America, Western Europe, Japan, and a few other places, access to wireless services is a commodity that most business users today would term a necessity rather than a luxury. The operators in such countries can count on a subscriber base that generates a reliable revenue stream through business use and an
14
GSM and Personal Communications Handbook
increasing contribution through private luxury use of cellular services. The notion of using mobile phones in the first place originated within tiny segments of the industrialized economies that used inefficient precellular radio systems for routine mobile communications. These systems have almost completely passed from existence today, because they were extraordinarily inefficient and expensive. Cellular systems today are efficient enough to attract and carry mass traffic at a reasonable cost. Modern marketing schemes, in general, practice in mass markets and invade cellular markets as they respond to and create different user patterns, primarily through tariffing policies. In many developing countries and regions (particularly the rapidly developing ones of South Korea, Taiwan, Hong Kong, Singapore, and, more recently, some Eastern European regions, China, and India) the conventional telecommunications infrastructure cannot keep pace with fast economic growth and the inherent requirement for reliable mobile access to basic telecommunications services that comes with economic development. The wired infrastructure is a very expensive machine that takes years to build and lots of money to maintain. Modern wireless technology is an attractive alternative that meets the demand for basic telephone services in developing economies. Cellular services and fixed WLL technologies kick in where time and money leave off. WLL will increasingly take over the role of copper access to residential and business customers, since wireless techniques provide low-cost access very quickly at a tiny fraction of the maintenance costs of fixed wire. The wired plant has a high fixed cost for its initial installation under streets and along poles. The cables need to be replaced over intervals of a few decades, and those who desire service demand a few meters of additional cable at the edges of the fixed network. This recurring expense is very high. The adoption of wireless has also been encouraged by deregulation and introduction of competition, a relatively new and potent combination in developing economies, and many developed ones as well. Wireless communications systems should be regarded as enablers of economic growth. This is why their development and installation are generally welcomed and supported in most of the potentially growing economies. Local telecommunications authorities and service providers join forces with experienced operators and investors (usually in the Western economies, though more are found in Asia today) in order to speed up the deployment of systems and services. Deregulation through the
The changing scene—again
15
granting of multiple licenses to multiple operators is another catalyst to rapid growth. The choice of equipment and standards in developing economies is devoid of philosophy and marketing hyperbole; it is based on availability, performance, and price. Because GSM is a mature technology, it is given preference in most regions. The system works, the equipment is available at relatively low cost, there is plenty of it, and the capability of offering international roaming with a multitude of other countries and operators is regarded as a political and social benefit. 1.2.2.2 Market implications for cellular and personal communications equipment and services
Whereas in Europe and Asia GSM-based systems are well represented, other markets, such as Japan and traditionally AMPS-based countries, are in favor of other competing digital cellular standards. Because the deployment of a TDMA overlay to AMPS in the United States was a disappointment in terms of market acceptance, and technical and commercial benefits to the operator were slow in coming, we can see some slowing in the acceptance of this technology exacerbated somewhat by CDMA claims. Still, the TDMA technology, particularly that represented by the IS-136 standard, is widely available and affordable, and will be more suitable than anything else in some cases. The PCS single-mode version for North America (IS-136) is one of the few low-risk choices for operators there. Cellular CDMA can be called a revolutionary technology based on a well-established methodology [4], which has been late to the market when compared to its digital rivals. CDMA cellular technology was proposed, developed, and thoroughly described by Qualcomm Inc. of San Diego. Dual-mode cellular with CDMA as the digital half, as proposed for North America in IS-95 with AMPS interworking, was also discussed in some form in Japan (interworking with analog TACS). Single-mode CDMA versions are going into operation in North American PCS bands and are likely to appear in Japan too. A number of factors and issues need to be considered by regulators (telecom ministries, commissions, etc.), operators, and investors when a choice for a particular network technology has to be made in light of the ystem quality required for the expected traffic mix and service requirements.
16
GSM and Personal Communications Handbook ◗ The cost of cellular infrastructure and terminals, operational costs, and typi-
cal performance: Open standards allow multiple vendor possibilities, which reduces costs through competition enabled by easy compatibility. This is valid to some degree with terminals. The infrastructure makers always practice some form of protectionism, even with GSM. Buyers tend to end up with one supplier for the infrastructure, even though complete compatibility of network components through definition in the standards was a declared target. To a certain degree, it is possible to draw a line between the base station subsystem (BSS) and the mobile services switching center (MSC) among different vendors. Infrastructure cost mainly determines, in combination with prospective subscriber figures and revenues, the break-even point (investment recovered) and the share value for the investors. ◗ Compatibility and ability to upgrade with existing equipment and dual-
mode operation (e.g., AMPS plus TDMA or AMPS plus CDMA): Operators who want to meet capacity needs in certain areas through the deployment of digital technology might want to reuse and upgrade existing (paid for) infrastructure for cost and compatibility reasons. The analog AMPS technology can serve as a fallback position when digital services are not available due to coverage or system loading. ◗ Availability: Even the best technology for a certain application is of
no use when it is not available for deployment (network technology) and mass production (terminals) at a scale sufficient to generate revenue for the operator. The GSM case against CDMA is an excellent example. Due to GSM’s 5-year head start and its wide support in the industry with a variety of IPR holders, many systems decisions were made in favor of this technology. Deployment of CDMA cellular service in the United States was postponed several times due to technical problems and the resulting lack of “ripe” network and subscriber equipment. Further delays also resulted from the fact that essential IPRs and first prototype products resided with only one industry player. Although there was wide support within the cellular industry for CDMA, some segments were made more cautious by the continued “hype” surrounding the technology even in the face of normal and expected technical delays. Once the implements for the technology are available, CDMA will get its
The changing scene—again
17
share of the market. This share could have been larger had the technology been made to work a few years earlier. ◗ Spectrum management: Because there are only chunks of radio spec-
trum available to operators and this resource is seen as scarce and precious, each of the operators needs to exploit what is available to the greatest extent possible. Various standards and technology implementations from different manufacturers allow different degrees of optimum spectrum usage and smooth capacity adjustments. From a system technology point of view, CDMA is probably the better choice because minimal frequency planning is required. Even though initial claims of spectrum efficiency (over AMPS, TDMA, and GSM) have been constantly reduced from “more than 40 times AMPS” to more realistic figures, CDMA/IS-95 is in a very good position here. TDMA-based technology gains some ground in spectrum management efficiency through clever cell-splitting schemes, microcells, picocells, umbrella cells, frequency reuse patterns, intelligent overlays, discontinuous transmission (DTX), and frequency hopping.
1.3 Aspects on marketing the product The transition from commercial or business use to a more consumeroriented market can be seen in many countries and for all system standards today. Different needs and demands, different technical solutions and applications, and different marketing requirements will accompany the migration of cellular services to a consumer commodity. The first users of cellular systems are those who specifically require mobile connections to the public-switched telephone network (PSTN). After a magic penetration threshold, say 10% (the number will be different for each country and market), is exceeded in a country, private users account for most of the additional growth. Relative cost of ownership (billing packages and terminal prices) as well as the perceived grade and value of the service are the key factors governing the transition beyond the threshold. Remember, these factors are to some degree operator driven (subsidizing phones, see Section 1.4.3) and can be a means for the operator to control usage. One example: Consumers use networks at different
18
GSM and Personal Communications Handbook
times than business users. This partitioning of use has an impact on system capacity and creates a new marketing playground for selling off-peak per-hour usage at lower prices. Operators have a certain freedom in controlling the factors of network usage against capacity by tariffing: high tariffs will not attract too many private users and may scare away some business users. This means that low usage and few subscribers can be a way to generate a sound revenue stream on a network that has not yet been built to full capacity. When more capacity is finally available, lower tariffs can attract the private users at the correct time. Competition in deregulated environments needs to be taken into account when making such rate calculations. Mixes of tariffs with different fixed monthly connection rates, and a variety of per-minute and per-second charges, allow a better match with customer expectations and behavior. The business user with typically hundreds of air-time minutes per month would rather choose a high fixed monthly rate with a lower per-minute charge. Some private users who would not use the phone much at all (tens of minutes a month) would rather pay a low monthly fee with a higher air-time charge. One other basic figure that underlines the shift in the cellular customer base profile is the worldwide average annual service revenues per subscriber, which are expected to drop from approximately $1,000 in 1995 to less than about $700 during 2000. Still, the total service revenue market in 1996 was worth more than $68 billion, and this figure is expected to grow to more than $220 billion by 2000 [2]. At the time of this writing, less than 3% of the world’s population had a mobile phone. In other words, 97% of the world’s population does not have a mobile phone—what a potential market! Technology becomes less of a marketing issue as rival standards deliver similar services at similar quality levels. Consequently, the religious battle between GSM and CDMA is better not fought in front of the end customer because she or he cares the least about the access scheme used on the radio interface or other technical babble. 1.3.1
Service providers
With the advent of deregulation and multiple network operators, cellular service providers appeared in abundance in many countries. This was particularly true for the introduction of GSM in Europe, when, for instance, in Germany 14 such organizations initially set out to collect a
The changing scene—again
19
fortune. However, the advent of service providers also has a regulatory touch and impact. Now, what is a service provider? A service provider is a reseller of air time. All other services offered through the service provider are arranged around the provisioning of air time. The network operator pays the service provider a commission for the air time sold to users. In other words, the service provider buys services from the network operators, which are then packaged for the user through adding some value. A service provider may market services from different networks at the same time. Differentiation among service providers is often made through different packages or “bundles” (which include the phone), convenient places (outlets) for people to look at phones, special service offerings, and unique features such as billing models and exchanges. Subsidizing of phones, which provides a low entrance threshold for a new subscriber, is made possible through long-term contracts (12 or 24 months). For new private network operators having no or only narrow distribution channels, the service provider bridges the gap between the network operator and the cellular customer. In this context, distribution means: ◗ Selling air time; ◗ Handling (selling) phones and accessories; ◗ Dealing with SIM cards (in the case of GSM); ◗ Handling repairs and exchanges; ◗ Handling subscriptions and billing; ◗ Customizing service features and billing models.
The emergence and development of value-adding service providers have resulted in many changes and innovations in the cellular business. These changes will continue. Many reorganizations, mergers, and acquisitions have taken place. Such a shake-out, in general, leads to fewer organizations with broader service portfolios. Even though many network operators continue to build their own distribution channels and in-house service provider organizations, service providers play an important role in the development of subscriber bases and the efficient provisioning of customer services and packages. Service providers depend on the service
20
GSM and Personal Communications Handbook
features offered by the operators and the interoperability of their networks. The network operator’s ability to offer constantly improving mobile telephone service of great quality enables the service providers to offer innovative and efficient customer services that attract users to the networks. 1.3.2
Fulfillment houses
The operator’s primary job is to provide reliable mobile connections to the PSTN and collect fees for the service. Getting phones out to new customers is a difficult and distracting task for many operators, which is why so many of them take advantage of service providers. Some operators prefer much more control and bypass service providers, but hesitate to become involved with the phones themselves. A fulfillment house can relieve the operator of the drudgery of getting phones to customers and still allow some measure of control. Fulfillment houses understand that the phone is the customer’s first and usually only vision of the network operator, and they are experts at getting phones out to customers quickly and correctly. Customers are invited, perhaps through TV ads, to call a toll-free number in order to arrange for delivery of a properly configured phone with which they can enjoy service. Most customers place their toll-free calls, make their billing arrangements, and receive their phones in the mail with a general feeling of efficient convenience. They think they are dealing directly with the operator and are seldom aware that an agent, a fulfillment house, has done all the work: taken the calls, warehoused the phones, tested and configured the phones (and SIM cards), packed the phones, shipped the phones, and collected the money.
1.4 Phones: shrink them, drop their price, and grow their features To the user, the image of wireless communications services is conveyed primarily through the terminal. The little tool that allows the user to communicate while on the move is the network. It can also serve as a status symbol, a gimmick for the technology freak, or a plain and common accessory that one has to have, just like a wristwatch. The evolution of mobile terminals in their technology and functionality was a fast one,
The changing scene—again
21
especially in the recent few years. It was directed by some old and basic requirements and some new ones too, but it was enabled by some important technological advances of the past 30 years. 1.4.1
What’s your size?
Since the introduction of cellular services, cellular phones for all frequency bands and for all standards have decreased in size. In the past 13 years they have fallen in weight from 2 kg to way below 200g and 100g as they shrank from liters (2,000 cm3) in volume to way below 0.2 liters (100 to 200 cm3). This trend continues. Japan is considered the benchmark with below 100-cm3 volume phones that have week-long standby times within the PDC and PHS systems. Heavy car-mounted units, a relic from the early days of cellular, have given way to pocket “Handies” and “Telefoninos.” Many new cellular phones are not intended for car use at all. However, in order to change a car into a mobile phone booth, one merely has to buy a car mount kit. This kit, however, may be expensive when compared to the cost of the plain phone, which is subsidized (see Section 1.4.3). The car kit includes a power supply for charging, a microphone and speaker for hands-free operation, and an external antenna. Using the “handy” (phone) with or even without all these appliances is not advisable in a car. Driving a car demands our full attention. The attentive and responsible telephonist on the move stops the car in order to complete a call. In some regions, it’s the law. Furthermore, in the case of the user who does not buy a car mount kit, radio transmission and reception with the built-in antenna are rather poor in the Faraday cages of cars. For operators who want to meet the anytime/anywhere expectation of their customers, the use of hand-portables from inside cars (without car kits) demands a more dense network with more base stations. Today, cellular mobile stations are increasingly worn rather than carried along, thus underlining the “personal communicator” concept. 1.4.2 How long can you stand by?
Battery-operated phones offer longer standby and talk times than ever before. Systems standards can optimize battery operation by setting functional standby requirements such as active and passive monitoring periods for paging. In addition, current semiconductor technology, together with an infrastructure built up with small cells for lower transmit power,
22
GSM and Personal Communications Handbook
provides for several hours of talk time and standby times in excess of 1 week. Early mobile stations that were not connected to a car battery, but were carried around with bulky battery packs, had talk times of nearly 1 hour and several hours of standby. Standby time greatly depends on the semiconductor technology used in the terminals. The standby current (measured in milliamps, mA) is a function of the time certain necessary entities within the terminal have to be active (duty cycles), the efficiency of the signal processing, and the silicon technology on which the products are implemented. Talk time depends, to a great degree, on the radio transmitter output power and transmit path power efficiency (power amplifier stage). The baseband sections that are active in conversation mode draw some considerable power from the battery and as such contribute to the discharge of the battery. 1.4.3
Ninety-nine cents?
Personal communication products have also become cheaper. Open standards, competitive markets, and other factors drive down the manufacturing cost of terminals by about 10% to 20% every year. But, watch out! Customer prices may not be what they appear. We all know the advertisements in which cellular phones are offered for an imaginative low price, say, “99 cents.” Such subsidizing practices are common now worldwide, but they distort the pricing structures. The term subscriber churn describes one negative effect arising from this marketing tool. Subscribers disconnect and even switch service providers or network operators after a binding period (say, 12 months) is over. They do not let their original operator recover the investment for the phone and earn a profit, thus tying up an increasing portion of the operator’s capital in phones wandering loose over the landscape. The subscriber is often attracted by a competing operator with lower tariffs and newer (and often better) phones that are subsidized again. The user’s only hassle is that a new mobile phone number has to be assigned with the new subscription. With less customer loyalty, operators need to revisit this practice and either decrease their sponsoring share on the phones, increase binding periods, or add features that increase the inconvenience of changing networks (e.g., e-mail addresses). New consumer tariffs and prepaid models in some countries add services and features that allow the operator to tailor the subscriptions to the user in such ways as to prevent churn.
The changing scene—again
23
1.4.4 What can you do that I can’t?
The increasing number of services offered by the mobile networks are supported in today’s terminals. Such features include short message services (SMS), supplementary services (SS, known from the ISDN), receiving and transmitting faxes (supported by the relevant protocols and connection to appropriate terminal equipment [TE] supporting fax; e.g., a notebook computer), data services including Internet capability (supported by appropriate TE connected to or integrated with the phone), and a growing list of others. Thanks to open standards, a variety of hardware and software is available today. The hardware and software accessories do not always come from the original equipment manufacturers; instead they come from an increasing number of third-party vendors. Again GSM, as the world’s most widely deployed digital standard, inherently offers an abundance of service features through its open definitions and standards. A number of features are found in many modern phones. Some of these features that have increasingly little to do with the communications task include real-time clock with alarm, phone books, hands-free (for car use) function, memo functions (making use of memory and voice compression), answering machine (also found as a “Mailbox” in the network), and computer games (like Tetris). New flavors of personal communicators merging a Personal Digital Assistant (PDA), a fax machine, and a data terminal (for Internet browsing) with a digital cellular phone get lots of attention in the market. The Nokia Model 9000 Personal Communicator was the first product showing the way. What can we expect for the future? Features like voice recognition (“call office” or “call mum”)—as introduced by the Philips Genie model in early 1997—handwriting recognition (already seen in PDAs), and active echo cancellation (as opposed to echo suppression; see Chapter 11) are being discussed, developed, and introduced. 1.4.5 Multiple bands and multiple modes
Someone can use a single familiar terminal, in which dozens of phone numbers are stored, within any GSM network when traveling on the road and place calls on a private Digital Enhanced Cordless Telephone (DECT) System when in the office. There is a slight difference between multimode phones and multiband phones. The merging of different wireless radio
24
GSM and Personal Communications Handbook
standards into one terminal has been discussed for many years, and the first products have finally appeared. Multiband refers to a terminal that can support a certain basic standard (say, GSM) but allows calls on different frequency bands employing the same protocol (GSM 900, GSM/DCS 1800, or PCS 1900). From a technology point of view, the implementation for GSM-based systems is relatively easy and cheap. The whole baseband section remains the same in multiband phones, while the radio section within the phone needs to be designed in such a way that more than one frequency band can be supported. This alone is still a challenging exercise. Multimode is a more general term that usually means a mix of a digital standard with a conventional analog capability, such as AMPS plus 800-MHz CDMA. The term also implies a multiband operation such as AMPS plus GSM in North America, where GSM can only appear in the 1900-MHz band in North America. Mixing modes in one phone is a bigger technical challenge. Some applications for multimode and multiband phones can be seen in a few cases. First, a GSM operator may be granted GSM 1800 spectrum in order to meet some capacity needs, such as in hot spots or in corporate environments. With dual-band phones the operator can benefit from a larger flexibility in spectrum and user management. GSM 900 might be used for wide-area applications and GSM 1800 for small cells and in-building services. Second, phones used for roaming between different networks (often on different frequencies) with the same number, such as between Europe (GSM 900) and in the United States (PCS 1900). Also, GSM 1800 network customers in the United Kingdom, for example, were lacking the possible roaming opportunities into neighboring European countries over the common GSM 900 networks until dual-band phones were available. Another option for the three examples might be to use the SIM card as the roaming instrument rather then the whole phone. For example, a European takes his SIM card out of his GSM phone on his trip to the United States and rents a PCS 1900 phone when he gets there. Fourth, pure North American applications use the AMPS system as the roaming vehicle. The currently specified AMPS-plus-TDMA and AMPS-plus-CDMA devices do the job in the 800-MHz band. In North America, and in some other countries as well, we already have dualmode terminals and services: D-AMPS (“D” for dual mode or digital)
The changing scene—again
25
appears in combinations of AMPS plus IS-54/136 (TDMA) or AMPS plus IS95 CDMA. Both of these dual-mode system standards offer digital services where they are available and support AMPS as a fallback position in all other places. The digital resources of the serving networks carry the additional capacity The objective to create more capacity in dual-mode networks can only be achieved if a certain high percentage of subscribers uses the dual-mode terminals. A more exotic single-band, triple-mode digital phone (TDMA/CDMA/GSM) would work in the American 1900MHz band, but may not have a market. Fifth, the exotic triple-mode device is not likely to appear in North America, but the triple-mode TDMA phone is already a reality: AMPS plus IS-136 in the 800-MHz band is added to digital-only IS-136 on the 1900-MHz band. A 1996 initiative of the GSM MoU (Memorandum of Understanding) network operator association called for a World GSM Phone (also called simply the World Phone) that would operate in all three frequency bands in which GSM is deployed, namely, 900, 1800, and 1900 MHz. The World GSM Phone would allow “global roaming” among all three system standards. Given GSM’s wide acceptance in so many countries, the “global” term applies. Other flavors could be GSM plus DECT, or all North American PCS digital standards combined with AMPS. For Japan, a combination of the personal handy phone (PHP) and the personal digital cellular (PDC) systems is possible. Even GSM/PHP is an option being discussed. Satellite communications technology will eventually appear, especially in areas where cellular or PCS coverage is not available. The dualmode option (satellite plus cellular) is an ideal complement in isolated regions, and remains a solution for filling in some global roaming holes. In addition to multimode satellite communication, features such as geographic location functions, which make use of the global positioning system (GPS), can be added to the terminals. The GSM (900/1800) plus DECT combination can be attractive for operators and users when both wide-area and local business locations need to be covered with one terminal. The subscriber’s home would be included with appropriate tariffs. The GSM/DECT combination is a flexible solution for outside salespeople and telecommuters who need reliable communications through one number. Multiple numbers might be used for one subscriber with only one communicator; different private and
26
GSM and Personal Communications Handbook
business numbers can presort incoming calls depending on the alert tone assigned to each of them. New licenses and spectrum allocations, deregulation, and the use of unlicensed bands (available to all operators and users) in combination with licensed bands (available only to particular operators and subscribers) may lead to different flavors of combinations and even hybrid systems in different areas worldwide. The success of such multimode technologies and their technical variations depends on market dynamics, competition, cost, and acceptance. The set of multiple-personality phones that will finally appear is much smaller than the number of different varieties we can imagine.
1.5 What is personal communications? The PCS buzzword is not new in the worldwide wireless industry. The PCS term, in particular, is used and misused in North America. Since 1995, the notion of personal communications was coupled with the words systems and services. In fact, we should go back to the end of the 1980s, when the first personal communications concepts were discussed in the United Kingdom, still one of the most competitive cellular markets in the world. Can we define what personal communications stands for? What will it be? Is it a new standard or a new network? What does a personal communications system offer that other systems do not? When we gather together the different definitions, advertisements, explanations, and technology offerings of the mobile marketplace, we find some common problems, requirements, design targets, and technical solutions. Changes in lifestyle and work habits have increased the mobility of people in recent years. People in industrialized countries are forced to increase their productivity by managing more information relative to things. Remote working and longer work hours have blurred the border between personal time and work time in the information society. The blending of work and home is enabled with supplements to plain old voice telephony. Now, many workers routinely add video, fax, and data communications to their voice communications tools. The process feeds on itself. Additional low-cost, non-voice communications services make it easier for employers to expect more performance out of their busy
The changing scene—again
27
employees, and for customers to expect even faster service from their suppliers. The communicating society finds itself using more and more valueadded services that did not exist a few years ago. Point-to-multipoint information systems such as cell broadcast, short messaging, and supplementary (personalized) paging services seem ideal for alerting people to sports scores, business events, weather, and stock price changes. Packet data services, in addition to conventional point-to-point telecommunications, add interactive features. The convergence of communications equipment and computers continues to create new products and new services for mobile users. The challenge is to make these products reliable and simple to use, but affordable. The cycle is an enticing trap that guarantees further innovation, for the very tools and services that force people to carry their work around with them are the same things that seem to offer a way out of the trap. Look at how things have escalated. Fifteen years ago an airmail letter was fast enough for most needs. Today, a fax is seldom fast enough. The more features and gadgets people can use with their phones, the greater will be the number of people who feel they cannot function without at least one of them. The picture is not yet clear. We have listed some features and services people often associate with PCS and other personal communications terms, but we still don’t know what it is. Let’s try a more orderly approach 1.5.1 PCS: defining the requirements
Boiling the requirements down to a few visionary buzzwords would mean to say: receive, revise, and originate calls or messages… ◗ With one small terminal; ◗ Everywhere (in the world); ◗ Under one number; ◗ In any form; ◗ At any time.
Communications and related services become personalized, customized, and location independent in PCS. We connect to people, not places. We dial
28
GSM and Personal Communications Handbook
one number to reach one person, regardless of where this person is. The communicator, and all the things the user does with it, becomes part of the person. The person and the output of her labor become ubiquitous, in both time and space, to a growing number of people. What processes and services make this happen? 1.5.1.1
Cost of ownership
Subscriptions have to be affordable to a wide variety of mass users. We are talking about individual and personal services for the mass market that compete with, and to some degree replace, the wired phone services. To be successful, even in developing countries, such services must be priced at levels matching the wealth structures of the customers. Service revenues per subscriber are shrinking, because the majority of new entrants to wireless network services are private users and not business users. Western Europe will see a drop in annual revenues per subscriber of approximately 30% from 1995 to the year 2000 [2]. Similar decreases in the range of 20% to 40% are to be expected in all other regions worldwide. 1.5.1.2
Access, mobility, connectivity, and services
Access to both wireless and fixed networks will be supported with mobility. The user wants to be able to originate and receive calls that are routed through public switches, to and from practically anywhere in the world, without any restrictions. Furthermore, in order to be reachable all the time and at any location (with the option of some private imitations), some kind of sophisticated mobility management has to be employed. This should be transparent to the user without the need to call in with a roaming code for the area in which the user is currently located. Users do not want to participate directly in the mobility management processes. Services and applications have to be state of the art and scalable to the user. Bearer services should allow for high data rates, up to 144 Kbps (basic rate ISDN) and fast packet oriented delivery, say, for fast Internet access. The portfolio is completed with voice mailboxes, messaging, and alerting—each of which has to function well when users do not need to be reached or prefer not to be disturbed. Combinations of business and private use, overlaid on even more combinations of office, residential, and wide-area use, challenge the system and its services. One terminal, one number (or explicitly two numbers: one for incoming private and another for business calls) will be
The changing scene—again
29
possible. Different billing and even different radio access may be provided for different applications of the same communicator. The concept is to have a single terminal for each user. The user can be sitting still at home or walking in the street, at work or at leisure, alone in the office or with a client in a meeting. The user does not have the time or the skill to assist in mobility management tasks or to deal with the details of his private branch exchange (PBX), roaming, and handover. 1.5.1.3
Coverage and capacity
Network coverage and capacity must be able to cope with a high user density, a broad mix of traffic, and sharp spikes in peak usage. Peaks will occur during daytime business use and evening private use. In conventional cellular networks this peak business capacity was cleverly exploited by operators by marketing the off-peak periods to the private user at “moonshine” tariffs. Adequate coverage to the private user means that she can use her phone anywhere in a given area she may find herself most of the time—say, 90% of her time. The service should be available everywhere and anytime: at home, on the road, on the street, in the office, outside, or inside. Roaming agreements between operators, or the fact that a particular operator’s service is nationwide, enhance the chances of being granted access to service all the time. 1.5.1.4
Voice quality
Voice quality has to be high. The trend is toward fixed-line toll quality without any compromises that are characteristic of first- and secondgeneration speech transcoder technologies. As a benchmark, the CCITT/ITU G721/726 adaptive differential pulse code modulation (ADPCM) algorithm delivers adequate quality, but at a relatively high rate of 32 Kbps. An abundance of speech coding algorithms is available that have been designed for use in communication systems with a requirement for low-bandwidth (baud rate) voice data, such as spectrally efficient wireless transmission systems, with few sacrifices in voice quality. Voice quality is determined by system design. Mobile systems dedicated to voice traffic can tolerate relatively high error rates on the channel relative to the lower error rates demanded of data services. Operators have to perform yet another balancing act as they configure their network densities in accordance with the mix of voice and data services their business plans require.
30
1.5.1.5
GSM and Personal Communications Handbook
The “communicators”
Phones have to be conveniently small, lightweight, secure, and easy to use. They must guarantee long standby times in excess of 1 week and talk times of several hours on each battery charge. There are practical and technical limits to these requirements. If one has to move an ultrasmall phone from ear to mouth in order to listen and speak in a quasi-halfduplex mode, a tangible limit of size has been exceeded, which, in this case, is “undercut”. Phones cannot be made as small as the technology allows. We still have to be able to use the phone with comfort. Small, low-power phones also need additional support from the network architecture. The network’s cell structure must be built densely enough to accommodate low RF power transmissions from the phone at the required grade of service. New data features that are supported by modem ports and larger displays need to be reliable and obvious in their use. Finally, security features, such as user authentication and encryption, must be supported without the hassle and confusion that usually accompany security measures. 1.5.2 PCS: the technical solutions to the requirements
Under the defined PCS circumstances, the basic demands are sufficiently met by a migration of current digital cellular technology into a technology that can handle higher capacity yet still allows for an upgrade path for services and applications. Existing digital technology has to be modified to accommodate more capacity while also being able to support additional services and applications. Cordless telephone-based technologies might be enough for some applications. DECT or PHP technology may work in dense office environments. Because nothing solved all the problems at the time, PCSs were initially regarded as “next” or “third”-generation systems. Apparently, the feature-rich second-generation digital cellular systems such as GSM are well suited for use in systems characterized as PCSs, but there is a lot of room for improvement in compatibility, worldwide coverage, cost, effectiveness against user expectations, and service features. Such improvements are currently tackled in projects defining and designing the next generation.
The changing scene—again 1.5.2.1
31
The cost issue
The economies of scale must provide for lower cost network installation and operation for the provider of PCSs so that the lower cost of ownership can filter down to the user. Terminals, let’s call them “personal communicators,” must be affordable without hesitancy. All this is only possible with agreement on open standards and competition in a large marketplace. GSM is the most widely accepted digital cellular standard and has the advantages of (1) being defined through a feature-rich open standard, (2) having a large number of technology backers, (3) having established a multiple-vendor environment with lots of equipment available at competitive prices, and (4) having embraced mass production not only of terminals but of infrastructure equipment that can be tailored to local requirements. Other standards, such as the North American TDMA (IS-136) standard, can challenge GSM in many areas, and may be chosen for such special reasons as network upgrade and rollout considerations. Other systems such as CDMA (IS-95) may eventually be seen as an alternative to GSM once they are mature enough to reduce some of the risks associated with new technologies. Cordless technologies like PHP and DECT are an alternative for some applications. Infrastructure investment costs per subscriber are in the area of several hundred dollars and somewhat less for cordless technology. There is plenty of room for improvement. 1.5.2.2
Access, mobility, connectivity, and services
Unlike the messy situation of 10 years ago, mobility and connectivity today are handled well with current cellular technology. GSM, for instance, offers well-defined structures and procedures for subscriber mobility and security [5,6]. The only limitation for mobility is outside a local network’s coverage area, and even that could be extended through roaming agreements with operators in other areas. These agreements can be made as invisible to the users as operators desire. The services offered through cellular and cordless technologies today are limited only by individual network and terminal capabilities. A large catalog of features is accessible to the personal communications user. Concepts introduced by the intelligent network (IN) approach in modern communications systems allow easy and flexible introductions of service facilities. INs separate the network intelligence from the physical
32
GSM and Personal Communications Handbook
switching and transport entities through defined hierarchies, protocols, and interfaces. The distinctions on whether access, transport, and termination of communication links are of a wired or a wireless nature, or are stationary or mobile, are completely transparent to users and to many network entities. Personal communications becomes part of the information superhighway concept and benefits from its intelligent structures. From high-quality plain voice service and enhanced mailbox functions to packet- and circuit-switched data and fax services, from Internet access and video over radio to paging and (personalized) short messaging, everything is possible. Why not send a video postcard over your GSM phone—sorry, through your communicator—back home to your parents, or to a desk neighbor in the office, such as was proposed and advertised by Ericsson [7]? Today’s cellular system standards define a multitude of services and service platforms that fit the PCS requirements for applications around wireless personal communications. GSM, through its Phase 2 features, is a best-in-class example for that. Today, the freedom offered to tailor the service offering to the needs and requirements of individual customers or user groups enables the personal communications market. Combinations of services and applications offered by different servicing networks, including the mobile network and the Internet, become a reality with a single terminal. 1.5.2.3
Planning for coverage and capacity
By making use of modern cell planning methods, cell splitting, directional antennas and sectored cells, dynamic power control, frequency hopping, and DTX, second-generation digital cellular systems are well suited to supply high capacities and provide good coverage within an area to be served at whatever quality of service is desired. If not for the financial investment and frequency planning effort, coverage and capacities could be increased by merely installing more base stations. One of the greatest challenges for all network operators is the acquisition and leasing of suitable sites, even for the tiny pico-sized base stations. Due to environmental, aesthetic, and health concerns, both imagined and real, more and more people refuse to allow radio transceivers to be installed in their neighborhoods. The human aspect is yet another parameter to be taken into account when planning a cellular radio network.
The changing scene—again
33
Cordless telephone systems, such as DECT, offer a dynamic channel allocation (DCA) technique that inherently makes best use of the available spectrum by only allocating channels that are tested and detected as unoccupied. Cordless technology, however, is best suited for low mobility and fixed wireless access applications. CDMA systems, through a very different approach, would work with far less frequency planning, but would still require thorough cell site and power planning [4]. It is also expected that CDMA systems may eventually provide a higher capacity for a given channel bandwidth, compared to TDMA techniques. A problem remains with increased loading of a CDMA channel; the transmission quality suffers and coverage decreases. Higher system capacities were traditionally achieved in cellular networks by cell splitting and intelligent, dynamic frequency allocations. With the migration to higher frequencies such as those used in PCNs or in PCS networks (1800 and 1900 MHz, respectively), more capacity can be achieved through a combination of some physical effects. Higher frequencies have a lower range, as the path loss increases with frequency. With the assignment of even wider spectrum (more channels) and shrinking relative bandwidths (channel bandwidths and center frequencies), and with the reduction of peak power levels in TDMA microcell systems for PCNs and PCSs, there is more freedom for network designers. With smaller cells and tighter radio transmission power control, better reuse of spectrum and more capacity can be achieved. Compare the few hundred meters to a few kilometers separation of GSM 1800 cell sites to the somewhat larger spacing that can be achieved with GSM at 900 MHz. This means that a minimum capacity network built just for coverage will have more spectrum reuse and thus more capacity (measured in Erlangs per megahertz per square kilometer; see below and Chapter 3) at 1800 MHz than at 900 MHz. However, as traffic demand builds, the 900MHz network can add extra cells to achieve exactly the same capacity. The propagation of radio waves and spectral efficiency depends on a number of parameters and has to be modeled by taking into account a number of conditions and variables. More theoretical treatment can be found in [8,9], in which additional references are given. There are a number of claims and scientific treatments from proponents of different technologies of what capacity can be achieved and at what cost. The comparisons are often expressed in Erlangs per square meter, or Erlangs per square kilometer, or (in the case of an in-building
34
GSM and Personal Communications Handbook
system) Erlangs per cubic meter. An Erlang is a dimensionless quantity used in statistical traffic studies. Traffic in Erlangs is equivalent to the number of calls carried in the circuits in 1 hour multiplied by the average duration of the calls in hours. One Erlang equals 3,600 call seconds per hour or 36 CCS (call century seconds) per hour. Expressed in this very general way, cordless telephone systems tend to have higher ratings than cellular systems. Cordless systems potentially have capacity ratings on the order of 600 to 10,000 Erlangs/km2, while cellular systems, as deployed today, have ratings in the low hundreds. This kind of comparison is handy in fixed-wire systems, but quickly loses utility in the study of mobile systems. Cordless and cellular systems, for example, are not designed to do the same things. Because the capacity of mobile networks is dominated by a host of other more complicated factors such as network latency, traffic mix, and the availability of cell sites, such comparisons become oversimplifications that confuse rather than enlighten. The proper approach is to make the best of the chosen technology using the tools appropriate to the selected technology. Full-service coverage may be achieved by dual-mode or dual-band systems and terminals, which are widely discussed in the industry and, in some cases, will eventually be introduced. A nationwide system, such as in North America where AMPS is the common interface, or a global system (GSM is the only example) would serve as the common denominator for a market. GSM 900 dual-band operation with GSM 1800 is part of ongoing specification work. A cordless standard supported in a dualmode phone may be able to fulfill the low-range office and home functions. There are proposals to even include low-earth-orbit (LEO) satellitebased mobile systems to supply seamless coverage. This, however, should be seen as part of third-generation systems. 1.5.2.4
Voice quality
High voice quality is achieved through the digital transmission of speech codec (vocoder) data. Sampled voice signals are compressed and error protected. A variety of powerful algorithms have been developed for wireless applications and progress continues on the lower rate possibilities. Though there is plenty of redundancy in the human voice, there is limited freedom in voice coding. The aim is the highest quality at the lowest rate possible. The price we pay for low data rate voice with good quality is complexity, including lots of processor resources (and power
The changing scene—again
35
consumption) and memory (cost). See Chapter 11 for a treatment of this subject. Most of the North American GSM systems feel a threat from the competing vocoder technologies used in CDMA systems. The GSM full-rate vocoder is substituted by the so-called “enhanced” full-rate (EFR) vocoder, both of which use the same compression rate (13 Kbps) and channel coding. But the EFR offers much better voice quality. The EFR demands more complexity in signal processing and more program and data memory than the older full rate vocoder. In the newer CDMA and TDMA systems, voice compression technology has progressed beyond that initially achieved in the original GSM proposals. Half-rate voice codecs have also been defined and proposed for TDMA systems like GSM The half-rate process uses every other TDMA frame (one time slot) for a single voice communication link, thus doubling the network capacity if all the users were halfrate users. The penalty is even more complexity with some sacrifice in voice transmission quality that falls behind the already low rated—in perceived quality—GSM full-rate vocoder. The current requirements for the next generations of vocoder technology are simple: lower rates (like GSM half-rate, or below) or variable rate, with toll speech quality (as good as that found with adaptive differential pulse code modulation (ADPCM) or the International Telecommunications Union [ITU; formerly CCITT] standard G.721). Also, in order to ease the requirements on network density, improved immunity to high bit error rates (BERs) is preferred. Complexity and memory requirements should be within the range of feasible technology. Some thoughts are also given to increased protection of lower rate signals through more powerful channel coding in order to make voice codecs more robust in marginal reception conditions. Additional channel coding would allow a greater mix of data serves onto the networks. 1.5.2.5
Communicators
The evolution in cellular terminal technology paved the way for personal communicators to become attractive toys and tools that meet a wide variety of consumer tastes and requirements. Today, mobile phones offer a high degree of security with user authentication and encryption. GSM’s subscriber identity module (SIM), which supports such features, is an excellent example. The advent of PCS can be seen as something that initiated a second evolution in product innovation as design cycles became shorter
36
GSM and Personal Communications Handbook
from generation to generation. The benefits of this second evolutionary phase will be fed back to traditional cellular applications, thus making the distinction between cellular and PCS harder to determine. The pager was, in fact, the first personal communicator. There are many differences between the old beeper, the alphanumeric pager, and the two-way messaging device. The trend follows cellular and personal communications systems: shrinking size, increased battery life, and a growing number of features. Modern paging services allow much more than plain pointto-point messaging. Much more customized information transfer supplements traditional point-to-multipoint service these days. GSM’s SMS-CBCH adds some familiar paging services and features to mobile phones. 1.5.3 PCS and what system technology?
The general requirements for PCS can be met by today’s secondgeneration digital cellular technology and its evolution, regardless of the frequency band in which it is deployed. From the service provision aspect, the technical details of which frequency band is used and which access technology and baseband algorithms are employed should be transparent and irrelevant to the PCS user. The borders between digital cellular and PCS started to blur even before the first PCS networks went into operation. The cellular operators started to argue that what they had been providing for some time was already personal communication services. It remains to be seen how the new PCS operators force the cellular operators to offer and market personal communications services, which they will do only when they have to. The personal communication “marketing trick” was first successfully applied to the PCNs in the United Kingdom. The networks launched GSM 1800–based service shortly after GSM 900, and competed with cellular in many areas, especially in densely populated regions, which were the initial focus for PCN deployments. PCS systems also reach out to replace fixed-line home access. This is possible even in industrialized countries, where fixed access is no commodity but a “given.” With good coverage, sufficient capacity, careful tariffing, and innovative service offerings, PCS operators and their alliances can get their share as long as they can attract new users. Left alone, cellular operators would have eventually adopted the personal communicator
The changing scene—again
37
and added PCS-like services to their networks. Competition from new operators licensed in new higher frequency bands forced the new operators to adopt something to distinguish themselves from cellular service: new phones, new features, and a new name—PCS. GSM and its little brother and sister, GSM 1800 and PCS 1900 (GSM-NA), with their abundance of service features and platforms, are in a good position to be deployed in many areas where a new mobile network needs to survive in a highly competitive market. Other digital cellular technologies must try hard to compete with the GSM world standard. 1.5.4
Where is next?
As is the case with most business sectors, forecasting future development is not easy in wireless personal communications. We can look at what is available today, what is ready to be deployed very soon, and what is on the drawing boards of the people tasked with merging future technology with consumer and operator requirements into third-generation system definitions and products. PCS and current digital cellular already offer a great deal. What else can be improved and what can be added? More bandwidth is one: more bandwidth for higher baud rate data (multimedia services and applications), bandwidth on demand, improved circuitswitched and connectionless packet data services, and better speech quality. Better integration is another: even higher integration of residential, office, and cellular services into a single terminal can be achieved. A user should be able to virtually carry along his home service environment or profile into visited networks when roaming. This is possible, for instance, with a personalized subscriber module similar to the GSM SIM. High system capacities and lots of spectrum are required far beyond what is typical today. What is needed? System definition work has started all over again in order to accommodate expanded requirements. Careful consideration has to be given to service definitions and system concepts, integration of improved mobility management, radio management (including direct satellite access and private branch radio access), and network management. The evolution of communications services and technology was not entirely self-driven. Some developments were put on track by regulatory bodies, concerted industry motivation, and standardization institutes. A lack of standardization leaves the industry awash in experiments that never end and high-priced products. Too much standardization squashes
38
GSM and Personal Communications Handbook
innovation. GSM has enjoyed a balance of influences. Europe, North America, and Asia reserve different research and development programs, which set the foundations for personal communications as we see it today, and as we are likely to see it tomorrow. The international approach to the search for a logical extension of PCS into and beyond the third-generations systems is found within the International Radio Consultative Committee (CCIR) of the ITU-R, a United Nations body, called the Future Public Land Mobile Telecommunications System (FPLMTS), which is now referred to as International Mobile Telecommunications 2000 System (IMT-2000). A detailed treatment of FPLMTS/IMT-2000 can be found in [10]. ITU’s standardization work on IMT-2000 aims to establish advanced third-generation global mobile communication services within the frequency bands identified by the World Administrative Radio Convention of 1992 (WARC’92). Spectrum has been set aside in the 1885- to 2025-MHz and 2110- to 2200-MHz bands. With different commercial and technical requirements in different areas of the world, we will not see a single system standard being implemented worldwide for IMT-2000. Rather, we have to look at IMT-2000 as a family concept for third-generation wireless systems and services that offer somewhat more than today’s secondgeneration systems.
1.5.4.1
Europe
In addition to spectrum allocations and operator licensing for the GSM 900 and GSM 1800 (PCN) bands, Europe is eagerly working on research and standardization for the communications scenario of the future. Programs in Europe include the RACE programs I and II (Research and Development of Advanced Communications Technologies in Europe) and the ACTS program (Advanced Communications Technologies and Services), with the final goal being to provide visions, definitions, and solutions for integrated broadband and radio telecommunication services [11]. The programs have also made contributions to third-generation mobile systems planned for the year 2000 and beyond. One such planned system is the European Universal Mobile Telecommunications System (UMTS). UMTS work is carried out in Europe by the European Telecommunications Standards Institute (ETSI). It is interesting to note that this initial work and responsibility were given to the Special Mobile Group (SMG), which is the group still in
The changing scene—again
39
charge of the GSM specifications. Input for this work is also provided by the 3rd Generation Interest Group (3GIG) within the GSM MoU. ETSI SMG is defining UMTS as the European third-generation system based on the FPLMTS/IMT-2000 framework, operating within frequencies reserved within WARC’92. UMTS standards and final ITU recommendations are expected to be available before the turn of the century, with system introduction in following years [12]. Still, many issues need to be resolved before that can happen. The technology and the service features need to be agreed on, and they have to be capable of providing voice, data, Internet access, and images/video anytime and anywhere. We speak about flexible data bandwidth from 150 Kbps (high mobility) up to and beyond 2 Mbps (low mobility). Regulatory issues need to be settled: who will license whom to do what in which market? There are many questions, as you can see. Eventually answers will be found and agreed on, and then new “next-generation” systems and services can become a reality. What is GSM’s role in this picture? The GSM standards, which are still reviewed and revised within ETSI, have been and will continue to be adapted and expanded with features and technology requirements. Tricky matters such as satellite interconnection support—as a complement to, rather than a competitor of cellular—have all kinds of dark passages and traps for the industry. Can satellite interconnection really be standardized, or would it be better to accept what is available from the two or three satellite operators who get their birds up first? Which approach will serve personal communications applications better? GSM is seen in the vision of UMTS as a reference feature set and was proposed to provide the evolution platform on which to build the future. Although it is not a pure “third-generation system,” GSM Phase 2+ standardization has a number of work items relating to UMTS. As such, UMTS is seen to provide somewhat more than was proposed by the more generic umbrella concept of IMT-2000. GSM and ISDN are also regarded as the core transport systems from which UMTS will have to evolve. GSM Phase 2+ comes close to the objectives of UMTS/IMT-2000 by covering broadband services for personal communication. However, UMTS/IMT-2000 requires a better integration of mobile and fixed networks and fully integrated, spectrum-efficient, and cost-efficient cordless telephony. The deployment of UMTS networks and services will realistically have to occur gradually. The standardization work will be carried
40
GSM and Personal Communications Handbook
out in phases. The first core specifications, including a radio interface standard (UTRA, UMTS, terrestrial radio access), are expected in 1998, and ETSI UMTS Phase 1 standards in the following year. Preoperational trials followed by real commercial operation would be possible in 2001 and 2002, respectively. This is expected to be the right timing in order to start alleviating the congested GSM networks expected at the time and to offer more multimedia-type services, which are beyond the scope and capabilities of GSM today. This would include unrestricted and fast mobile access to the Internet and intranets. Due to the required heavy investments and the complexity of yet another system rollout, a gradual transition employing dual-mode terminal equipment (GSM and UMTS)—ideally including seamless handover—is anticipated. In the terminal arena we are only limited by our own fantasies when trying to predict the flavors and combinations of technologies and applications that will emerge. The WARC’92 spectrum in which UMTS services may be deployed in Europe appears in the following bands: Cellular:
2 × 60 MHz paired (1920–1980 + 2110–2170 MHz)
Cordless asymmetric:
35 MHz unpaired (1900–1920 + 2010–2025 MHz)
Satellite:
2 × 30 MHz paired (1980–2010 + 2170–2200 MHz)
The total of 155 MHz of terrestrial spectrum has already been declared within the European wireless industry as far too little to support technically adequate and commercially viable services in mature (mass market) UMTS networks. Requests for as much as four times the spectrum were made. For comparison Europe has allocated 2 × 105 MHz of spectrum for GSM 900 and 1800 shared by up to four operators, and 20 MHz for DECT. Apart from UMTS, the GSM platform is still the model whenever definitions or standardization for mobile networks is considered for new systems and applications. An example of a new application is the collection of LEO satellite systems. An easier convergence of systems and services between different networks is anticipated through LEOs that can work across equal and common network platforms. The technical realization of the definitions and standards is left to the industry and their acceptance by the consumer. Timing the
The changing scene—again
41
standardization work (avoiding delays) and the introduction of appropriate services is critical to success. 1.5.4.2
North America
The North American approach, which, due to the large size of its economy, is driven by the United States, is somewhat different from Europe. The term PCS originated in the United States, and under this buzzword the personal communication systems and services that will take the North American continent into the next century were discussed, refined, and marketed (see also Section 1.5.5). PCS evolved from second-generation system technology and can be seen as the next logical step toward the universal personal telecommunications (UPT) vision building on platforms emerging from and interconnecting to PCSs [13]. Such platforms are either proposed or are already available within the industry, in accordance with definitions and requirements set up by various industry committees. The regulation of PCS and UPT is confined to spectrum assignments by the Federal Communications Commission (FCC) and the evaluation and recommendation of technology standards by the Joint Technical Committee (JTC) of the Telecommunications Industry Association (TIA) and the T1 committee. Eventually four physical layer radio interface standards were approved for PCS use in the newly assigned frequency band in North America. The four systems are PCS 1900 (GSM-NA), CDMAOne based on Interim Standard number 95 (IS-95), TDMA based on IS-136, and a composite CDMA/TDMA/FDMA air interface proposed by Omnipoint (IS-661). A few other proposals appeared as the systems got ready to deploy. The GSM derivative is the PCS 1900 system. The remaining three systems are contrasted with GSM in Chapter 3. With emerging PCS networks and with the amount of spectrum (part of FPLMTS frequencies from WARC’92) and licenses allocated, North America is likely to approach third-generation systems with a certain amount of attention paid to protecting the freedom of the self-regulating market and its own industry. This means that CDMA technology evolving from the narrowband IS-95 standard will be particularly favored. Backward compatibility with existing North American network and switching technology, including the installed cellular and PCS systems, will allow upgrades and operator flexibility for wideband services and new feature
42
GSM and Personal Communications Handbook
offerings. Adoption of a single North American wideband standard outside North America, for example, in parts of Asia and other continents, may be possible due to the strong foothold of North American wireless technology companies in other parts of the world. 1.5.4.3
Japan and Asia
In Japan the deregulation of cellular services and the introduction of PDC and PHP led to an extraordinary breakthrough of personal mobile communications. It also led to an extremely urgent need for spectrum and capacity. This need is driving today’s activities in the Japanese wireless industry and operator community. The Ministry of Posts and Telecommunications (MPT), NTT, and the industry-based Research and Development Center for Radio Systems (RCR) are the main technological drivers behind the new cellular developments. Along with Japan and South Korea, many other Asian countries are engaged with ITU-R and make their own contributions to the definitions of third-generation mobile systems, such as IMT-2000. With European GSM becoming a world standard, the Asian wireless communications community—more than North America’s—has lost out on catching the wave. A new chance is now seen in actively pushing and contributing to next-generation system standardization. It remains to be seen whether the results will match requirements and expectations, what compromises will be made, and what the regulatory and standardization policies will look like throughout the world. As mentioned earlier, different interests in different areas and regions already demonstrate that agreements (e.g., with European standards) on basic system technology such as a common air interface (CAI) standard or a common network protocol are not likely to be reached. National and regional interests and schedules have already led to the establishment of different positions and specifications for inputs into IMT-2000 standards. For example, the likely proposals for the CAI are ranging from various wideband CDMA to wideband TDMA and other access methods. Also, a balance needs to be found between short-term necessity, long-term requirements, and technical and commercial feasibility. 1.5.5 GSM and PCS in the United States: an overview
PCSs in the United States have had an interesting past, and their development continues to be the basis of table conversations, both formal
The changing scene—again
43
and informal, throughout land mobile industry. There were numerous clashes among the supporters of many different proposals for system implementations. The sale of frequency bands at extraordinary prices within a hands-off regulatory environment tested technical and nontechnical aspects of the proposals that were never seen before. There was plenty of confusion and suspense within the mobile radio industry, piles of money were at stake, and everyone learned many new lessons. 1.5.5.1 Personal communications in the United States: the potential market
The exploitation of the enormous potential of personal communications has become a major goal of today’s telecommunication industry. Existing old analog cellular networks in the United States offer a limited number of telecommunication services and have generally tolerable voice quality with severe capacity constraints in only the largest markets. The AMPS network is, however, stable and too expensive to simply throw away; it took 12 years and $27 billion to build [14]. The cellular operators had already signed up all the available customers, and no number of clever marketing tricks and additional phone subsidies were going to maintain the subscriber growth rates on which the wireless industry had grown to depend. The American operators needed new customers. It is obvious that PCS systems had to offer more in order to be successful in the mass market, because the mass market expects wireline quality voice and no connection difficulties at all. Even if the expensive tariffs were reduced to attract new consumer-type users, the largest cellular networks could not handle all the new traffic. The demand for better quality, more attractive services, and more capacity meant more spectrum had to be allocated. Personal communication wireless networks are meant to cover that need, and they can fill the need as either complements or as a competition to cellular. Cellular radio in the United States as of early 1997 claims about 40 million subscribers, which is a 13.5% penetration. Only a tiny fraction of these subscribers are on the new PCS networks, which are very new indeed. Predictions for the future cover a wide range of possibilities. It seems clear that the penetration will increase to about 35% by 2000 (which would be 100 million subscribers), and if the recent history of such predictions is any indication for the future, the 100 million figure is probably too low. It is, however, unlikely that most of the optimistic
44
GSM and Personal Communications Handbook
predictions for PCS’s share of this growth will actually come true. The most optimistic portion for PCS the authors can justify is about onequarter of the 100 million subscribers by 2000, and probably a bit less. 1.5.5.2
The PCS industry: the actions and the auctions
The “official” perspective for PCS in the United States was that the technology behind personal communications deployments could be anything. Whatever technology was selected, it needed to provide the required services at a certain quality and quantity level. Such a Wild West approach, it was argued, would leave lots of room for innovative equipment manufacturers and modern telecommunications service providers, thus optimizing the systems to the user’s advantage. Another aim was also to provide a fair and common ground for competition, involving as many relevant market forces as possible. This includes all the manufacturers of mobile communication technology, operators, regulatory bodies, and the users. “Let the market decide” was the slogan. A number of regulatory issues had to be addressed. The services offered by the PCSs had to be listed, spectrum allocations for licensed and unlicensed use had to be sorted out, spectrum licensing (for licensed operation) had to be organized, a map of geographic service areas for licensing had to be drawn, and standards needed to be reviewed. Actions for PCS started as early as 1989 with the FCC’s “ruling over public frequency allocations” being the guide for regulatory and legislative initiatives [13]. One of the most important outcomes of these activities was the allocation of new spectrum for PCS in September 1993, and the decision to make large chunks of spectrum, confined to designated service areas, available to a limited number of operators through auctions. In addition, spectrum for low-powered unlicensed operation was allocated. All the allocations had taken account of prospective technical needs such as bandwidth and geographic coverage. The PCSs had to provide their services through a complicated matrix of frequency allocations and geographic areas, called blocks, within the 1850- to 1990-MHz band. The frequency bands have various sizes to accommodate different modulation techniques and to effect spectrum sharing with different kinds of incumbent fixed point-to-point microwave users. The frequency bands are further divided over the entire country into trading areas. The two types of trading areas were originally derived from the 1992 Rand McNally
The changing scene—again
45
Commercial Atlas and Marketing Guide (123rd edition): (1) a Basic Trading Area (BTA) typically is an area around a city or other business area; (2) a Major Trading Area (MTA) roughly corresponds to the size of a state and consists of clusters of BTAs that are more or less economically independent. Only two operators are allowed in each MTA, and as many as five operators are allowed in each BTA. The largest blocks, the ones with the greatest population, are the A and B blocks, which are the MTAs. Smaller blocks are reserved for the BTAs, and have C, D, E, and F designations. The MTAs cover the BTAs. The blocks were sold to bidders during FCC auctions held in 1995 and 1996. We will not delve into the mountain of details except to say that they were far from hometown cattle auctions. Licenses for blocks A and B (the MTAs) raised more than $7.7 billion for the U.S. treasury. The winners of the bids for the 51 MTAs and the 493 BTAs are free to decide which technology and services they want to deploy. In a separate ruling, some narrowband PCS licenses were issued in the 900-MHz band during this time. Chapter 3 explains the details of the North America PCS frequency band allocations in the context of the different types of technologies selected by the operators. 1.5.5.3
Recovering the investments
Investment recovery, reasonable profit, and increased share value are only possible if service revenue is generated. This will not be easy for most of the American PCS operators. After major money is spent at auction for the blocks, radio spectrum used by microwave links may need to be cleared. Microwave users have a time frame of 3 to 5 years to migrate to new frequencies. Then, a radio system technology has to be chosen. The lowest risk selection is a technology that is proven to work, is available, and gives the users the required services at the expected quality. Operators who select low-risk, proven technologies expect the industry to be able to provide the required high-quality equipment in sufficient volume and on time. The prudent operator carries out all of these kinds of negotiations and plans before the licenses are awarded. The next step is to design the network and install the infrastructure. The network slowly comes to life as long as the bills are paid. The best efforts in network planning are compromised by site acquisition difficulties; insurmountable adversities drive the operator to second and third choice sites. Constant testing, tuning, and network design changes keep the engineering staff busy over long hours even when the equipment is installed “off the shelf.”
46
GSM and Personal Communications Handbook
Some operators supplement their own technical staffs with “friendly customers” or “early users” who are rather sophisticated, technically oriented customers tolerant of technical problems. These “customers” are heavy-duty users who, in return for a free phone and reduced connection charges, can be counted on to drive around and report all kinds of problems without the public relations overhead associated with typical paying customers. Eventually the time comes to start advertising and recruit as many subscribers as possible. Where do the customers come from? They may come from other PCS operators in the area, existing 800-MHz cellular customers, or—of course—new customers from the general public. The challenge for the American PCS operators is that most of the American population that needs a mobile phone already has one. Taking customers away from other wireless operators without incurring high sales costs and subsidies requires sophisticated marketing skills. New customers come from the mass of those who want a phone but may not need one. Most of these kinds of new users will not be the high-volume users the cellular operators built their fortunes on during the past 15 years, and others may have credit problems. The PCS operators will need to distinguish themselves from established wireless services with their PCS features. They will need to adapt services and tariffs constantly to remain competitive. Once the customers are won, they need to take extraordinary measures to prevent churn. Operators have to establish a high level of service for their best users, maintain it, and make constant improvements. The auditions are done, the bills are paid, the cast for the play is complete, some rehearsals (trials) have been performed, and the first scenes of the play have already begun. As services become available the dreaded shake-out begins; only a few of the operators will survive. The main characters have already armed themselves for the big fights; they have found allies and planned some conquests. 1.5.5.4
The standards—the challenge
The FCC did not mandate any standards to be used in the new allocated spectrum. Still, some basic rules about spectrum usage were stated, and recommendations on spectral etiquette were adopted [13]. The FCC presumed that spectrally efficient radio techniques could be accommodated in the smaller frequency allocations and reserved the larger ones for broadband services.
The changing scene—again
47
Due to adoption of wide open free market forces and deregulation, the industry (TIA/JTC) attempted to lend order to the chaos as it created a choice of seven standards (TAG-1 through TAG-7), which were named after the Technical Ad-hoc Groups that developed them. Among them we find wideband and narrowband CDMA systems, TDMA systems (both IS-136 and GSM), as well as DECT and PHP related approaches. They were all backed by their own supporters, and some even had potential operators. Seven standards also means seven different radio interface approaches. All of the proposed systems employ digital modulation and speech transmission techniques. There were no analog CAI proposals. Table 1.1 shows a list of the seven PCS standards proposals, together with some technical data for comparison. Of the seven standards in the table, only three have lived to be deployed: IS-95 (CDMA), IS-136 (TDMA), and GSM (called PCS 1900 or GSM-NA in North America). Motorola recently proposed a 1900-MHz version of their analog IS-88 (N-AMPS) system, and the PACS and IS-661 systems may eventually find a place in specialized settings. The American PCS operators have taken on a daunting task. They have to build systems and attract customers without the time and other advantages the original Table 1.1
List of American PCS Standards Proposals TAG-1
TAG-2
TAG-3
TAG-4
TAG-5
TAG-6
TAG-7
Standard derived from
new
IS-95
PACS
IS-136
DCS/GSM
DECT
(IS-661)
Access method
CDMA/ TDMA/ FDMA
DSCDMA
TDMA
TDMA
TDMA
TDMA
DSCDMA
Duplex method
TDD
FDD
TDD/ FDD
TDD/ FDD
TDD/ FDD
TDD
FDD
Modulation
QCPM
OQPSK / QPSK
π/4DQPSK
π/4DQPSK
GMSK
GFSK
OQPSK/ QPSK
Net bit rate (speech)
32 Kbps
8 and 13.3 Kbps
32 Kbps
7.95 Kbps
13 Kbps
32 Kbps
32 Kbps
Channel spacing
5 MHz
1.25 MHz 300 kHz
30 kHz
200 kHz
1728 kHz / 5 MHz 1250 kHz
Number and length of time slots
32 /
—
8/
6 (3) /
8/
32 /
312.5 µs
6.7 ms
577 µs
417 µs
625 µs
—
48
GSM and Personal Communications Handbook
cellular operators enjoyed, and they have to accomplish this feat in a much more competitive market. Only a few can survive. There is a maximum of only two cellular operators in the 800-MHz bands, and some areas have only one. Both the A-band and B-band cellular operators have spent billions of dollars over more than a decade building their AMPS systems, which, today, are linked with mature roaming agreements and interworking network protocols (IS-41). The PCS operators, in some cases, have to build systems from scratch, and there can be as many as five additional competitors as they spend their money. Who will the survivors be? The IS-136 proposal is championed by AT&T, which is putting much pressure on the wireless industry to supply triple-mode phones: (1) 800MHz AMPS, (2) 800-MHz TDMA, and (3) 1900-MHz TDMA. AT&T has the resources, the marketing skill, and the brand recognition to make the IS-136 protocol succeed. With the triple-mode phone, AT&T will enjoy the advantage of not having to build an entire national network. Two CDMA operators, notably PrimeCo and Sprint, need to build entire national networks unless dual-band, dual-mode (800-MHz AMPS plus 1900-MHz CDMA) phones appear at competitive prices. It is not clear when this will happen. Given the equipment shortages and continuing technical problems, both of these CDMA operators have done remarkable jobs building their networks in late 1996 and early 1997. The first PCS operators to offer services to the American public were GSM operators. This was due to the relatively easy availability of GSMbased equipment. The success of the American GSM operators is studied in Chapter 3 in light of their IS-136 and CDMA competitors. Though it is too early to judge how the competing technologies will sort themselves out in the market, the struggles are an interesting study of the relationship of competing technologies with money, marketing skills, and luck.
References [1]
Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chaps. 1 and 2.
[2]
Market Trends—Cellular Services Worldwide, Dataquest 1996.
[3]
PHS International, www://phsi.com.
The changing scene—again
49
[4] Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 13. [5] Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 7. [6] Mouly, M., and B. Pautet, The GSM System for Mobile Communications, Palaiseau, 1992, Chap. 7. [7] Ericsson GSM, http://www.ericsson.se. [8] Balston, D. M., and R. C. V. Macario (eds.), Cellular Radio Systems, Norwood, MA: Artech House, 1993, Chap. 1. [9] Turkmani, A., “The Mobile Radio Channel,” Chap. 3 in Personal Communications Systems and Technologies, J. Gardiner, and B. West (eds.), Norwood, MA: Artech House, 1995. [10] Fudge, R., and J. Gardiner, “The Future: Third-Generation Mobile Systems,” Chap. 12 in Personal Communications Systems and Technologies, J. Gardiner, and B. West (eds.), Norwood, MA: Artech House, 1995. [11] Gardiner, J., and B. West, “The Needs and Expectations of the Customer,” Chap. 1 in Personal Communications Systems and Technologies, J. Gardiner, and B. West (eds.), Norwood, MA: Artech House, 1995. [12] Rapeli, J. (Chairperson, ETSI Sub Technical Committee Universal Mobile Telecommunication System, SMG 5), Standardization for Global Mobile Communications in the 21st Century, http://www.etsi.fr. [13] Russel, J., and A. Kripalani, “A North American Perspective,” Chap. 10 in Personal Communications Systems and Technologies, J. Gardiner, and B. West (eds.), Norwood, MA: Artech House, 1995. [14] Cena, A. M., and K. O. Nielsen, Telecommunications Equipment, New York: Bear, Stearnes & Co., December 1996.
CHAPTER
2 Contents 2.1 GSM: what it was meant to be and what it became 2.2 The role of the GSM MoU 2.3 ETSI and the special mobile group
From Pan-European mobile telephone to global system for mobile communications
2.4 Standards: the present and the future 2.5 GSM type approval issues
I
f we dig up the roots of GSM, if we look back to the days when Europe’s cellular landscape was a patchwork of sundry incompatible analog mobile telephone systems, then we can understand the initial motivation and goals for a common Pan-European cellular standard [1]. Now, after we have passed a few standards and specifications milestones and tested and adopted the proposals in Europe and tried them out on the mobile communicating public in some countries beyond Europe’s traditional borders, we see that something has changed. We see that the vision of the early proponents of GSM has been utterly transformed to a global reality that in some places is not even referred to as cellular. What were the initial goals and the early experiences with GSM? When and why
51
52
GSM and Personal Communications Handbook
were certain services and features introduced? What were the roles of ETSI and the GSM MoU in GSM’s transformation, and how are they organized to meet their goals? What about personal communications networks (PCNs) and DCS 1800, and how did PCS 1900 appear in North America? As the GSM success story continues to be written, and as system standards continue to evolve to meet new requirements, we can revisit the original objectives and follow the changes to discover GSM’s place in the new world of PCSs and PCNs.
2.1 GSM: what it was meant to be and what it became 2.1.1
The initial goals of GSM
Once upon a time…. The original humble objectives of the Groupe Spécial Mobile, the working party established by the Conférence Européenne des Administrations des Postes et des Télécommunications (CEPT) in 1982, and the European communications community was to define a new common standard for mobile phones in Europe. The common standard was to be implemented in each of the original 12 European signatory countries, and it would be a common open standard throughout Europe that would allow international roaming and compatibility among many equipment vendors so as to reduce prices and facilitate easy interworking. Moreover, GSM would offer a wide range of attractive new services, features, and applications unknown in the older analog systems, and those old systems would—by decree—eventually be replaced by GSM. GSM would create a common market of attractive and competitive dimensions new to the European markets. Some milestones in GSM’s short history are listed in Table 2.1 [1]. 2.1.2
The initial results
What was achieved through this common work and cooperative effort within the European wireless community? These were the initial results: ◗ A complete cellular system standard, not just the air interface, was
written, describing wireless extensions to plain telephony that used
From Pan-European mobile telephone to GSM
53
Table 2.1
Milestones in the History of GSM Year
Milestone
1982
Groupe Spécial Mobile established by CEPT
1986
Permanent nucleus established
1987
GSM MoU group formed
1989 and ongoing
GSM becomes a technical committee (TC) within the recently founded European Telecommunications Standards Institute (ETSI); names change: Groupe Spécial Mobile becomes Special Mobile Group (SMG) within ETSI, and GSM stands for Global System for Mobile Communications
1990
Start of work on DCS 1800 based on GSM 900 Phase 1 specifications
1992
Commercial launch of GSM services in European networks
1993
Start of commercial DCS 1800 services (UK)
1994
First data services over GSM
1995
GSM Phase 2 core standards completed ETSI continues to work on Phase 2+ features PCS 1900 standardization within ANSI T1P1 in the United States 117 GSM/DCS/PCS networks on the air in 69 areas worldwide Approximately 50,000 cell sites installed worldwide
1996
Start of commercial PCS 1900 services in North America, 6 networks/operators, 17 MTAs/18 BTAs, with about 200,000 subscribers by year’s end More than 30 million subscribers in GSM networks; nearly 20 million new subscribers added in 1 year 175 networks on the air in 92 areas 215 network operators from 108 countries or territories are committed to GSM technology
1997
More than 15 new DCS 1800 networks and additional North American PCS networks go on air The first dual-band GSM/DCS handsets appear on the market GSM 1900 systems come up in smaller North American (C-block) markets More than 600,000 GSM subscribers in North American PCS World total GSM subscriber number by year’s end is around 55 million in 200 networks/109 countries
digital representations and transmission of coded voice and user rate data with a TDMA radio interface, and a close compatibility to ISDN services, signaling, and network architectures.
54
GSM and Personal Communications Handbook ◗ System specifications included a variety of services and application
platforms as well as advanced security features such as encryption and user authentication. ◗ A phased introduction of the services was implemented that accom-
modated itself to realistic standardization and productivity schedules, and recognized the limited availability of system components [1]. ◗ A tremendous level of support and interest from the whole manu-
facturing industry and prospective network operators was demonstrated even in countries not listed among the early signatories and supporters. The support was fueled by its own heat. The operators saw the chance to supply state-of-the art services with a relatively low initial investment and operating cost, and the manufacturers saw a huge market into which they could offer premium products at low cost. Today we see a living system standard that has come alive in many countries and is still evolving to accommodate the services demanded by the diverse cultures that have adopted GSM. 2.1.3
First experiences
GSM 900 specifications for Phase 1 were frozen in 1990, and the first networks started commercial service in 1992. Initial difficulties with network coverage, terminal type approval, handset shortages, echo, and some control software problems were gradually overcome, and Phase 1 GSM technology became stable in a short time. The matrix of roaming agreements between GSM network operators quickly filled up, and the quality of service was accepted by the subscribers. The early success of GSM cannot be wholly attributed to the standard and the products alone, because some countries and regions in which GSM was deployed were desperate for any kind of new cellular service. The availability of more bandwidth in the 900- and 1800-MHz bands, and the additional cellular radio capacity the new frequency allocations allowed, was a positive market driver in the United Kingdom, Germany, Italy, Portugal, and France. This was true even though, in some areas, GSM networks had to share spectrum with existing analog networks. The introduction of Phase 1 services was initially restricted to plain voice telephony. Other features such as fax/data, voice mail, short
From Pan-European mobile telephone to GSM
55
messaging, and call forwarding/blocking were not available in most networks. Mobile data and fax only became a commodity to businesspeople on the move when GSM PCMCIA cards that allow GSM phones to work with portable computers finally became available. The initial acceptance of data services, some of which had been available in many GSM networks since the start of 1995, was disappointing. According to the Mobile Data Initiative (MDI)—a cross-industry initiative aimed at promoting the use of data services [2]—data services in GSM networks accounted for only 0.5% of the traffic in 1995. This is very different from the situation in wired networks in which 46% of business traffic is data. As of 1995, data generated only 1% of the revenue in European GSM networks, and 1 out of 50 European GSM mobile phone users had a connecting device (PCMCIA card or special cable) for a mobile computer (notebook or laptop). The GSM specifications describe a multitude of data bearer services from circuit switched to packet switched, both connectionless and connection oriented, transparent and nontransparent, asynchronous and synchronous—all of them at different rates. With the advent of more application features and tools such as professional mobile data equipment, Internet browsers, and e-mail through PDAs, data services will, it is hoped, command a more respectable share of air time and revenue. Not all of the specified services may be present in a particular GSM network, and some of the applications are not particularly easy to use; the marketplace will determine which ones are the most popular and which ones will attract the applications and accessories. The GSM standards furnish the platforms for the applications, and the evolving GSM standards (GSM Phase 2+; see later discussion) enhance the performance of the applications through higher data rates, new services, and even newer network features. It is up to the terminal manufacturers to design easier and more familiar user interfaces, and it is up to the operators to make new features and data services attractive and accessible.
2.1.4
PCNs and DCS 1800
PCN, the first conscious approach to personal communications, was a U.K. initiative, triggered by the Department of Trade and Industry (DTI), and was perceived as an essential complement and competitor to cellular. Spectrum was available at 1800 MHz, and GSM was selected over, for
56
GSM and Personal Communications Handbook
example, cordless techniques as the most appropriate technology to support the new services. One important reason for choosing an established standard (GSM 900) over the alternative of developing a completely new system for PCN was early and easy availability of low-cost network equipment and terminals. With all the air interface details worked out and tested, it was seen as a plug-and-play solution to establishing a PCN. A proposal for developing a completely new standard would have also led to less incentive for the industry to invest in the development of new equipment and new standards. The main difference between GSM and PCN is in the amount of radio resources; the DCS 1800 system has 375 channels compared to GSM 900’s 175 channels (including 50 channels in the extended frequency band). The signaling differs chiefly in its ability to accommodate the channel numbering and reduced power level assignments. What impact did DCS 1800 have on the GSM specifications? When the GSM Phase 1 specifications were frozen, DCS was merely an add-on to GSM 900. Because the major difference was in the radio spectrum allocations, the specifications were edited as separate documents and the DCS prefix was substituted for the GSM designation. Specifications for digital cellular system 1800 (DCS 1800) were born within ETSI in 1990 when adaptations of GSM Phase 1 documents (delta documents) were published. Almost every essential part of the GSM 900 standards was kept; changes were confined almost entirely to those required to accommodate the new RF band, and only individual pages were edited in the protocol definitions. With the advent of GSM Phase 2, DCS 1800 became an integral part of the GSM specifications. The DCS designation was replaced by GSM, thus GSM 1800 instead of DCS 1800. Because the older DCS 1800 designation has been around for 6 years, and most of the literature refers to this form of GSM as such, we use the DCS 1800 reference in this book. The history of the PCNs and a more exhaustive treatment of the subject can be found in [3]. The first PCN services that used DCS 1800 technology started in late 1993 and early 1994. Networks opened up in the United Kingdom (One2-One and Orange) and in Germany (E-plus); others followed in Europe (France, Sweden, Switzerland, Greece) and Asia (Singapore, Thailand, Malaysia, Hong Kong). Though there were many field trials and experiments, full-scale cellular service had never before been deployed in the PCN frequency bands. Different propagation characteristics were
From Pan-European mobile telephone to GSM
57
observed with the limited radio range and lower power class mobiles in DCS 1800 networks (1W as opposed to 2W GSM handhelds), and the much greater spectrum allocation (three times over GSM 900) led to new network planning techniques different from those already known in analog and digital cellular below 1 GHz; that is, much higher capacities were achieved. New mobile PCNs often start out against existing, fully constructed, and thoroughly tested analog and digital cellular competition with nationwide coverage. How do the new operators attract and keep their customers? The customers are found in consumer mass markets, perhaps among those who have never owned a cellular phone, and in small regional enterprises, which may not need the wide roaming capability of GSM 900 and mature analog systems. Marketing efforts and the pricing of PCN services is tuned to these clients:
1. Lower cost service compared to conventional digital and analog services; 2. Quick activation—“walk out and call;” 3. Newer, state-of-the-art, and attractive terminals only available when advanced, second-generation PCNs appeared. It was left to the U.K. markets to sort out the early prospects for the survival of the DCS 1800 operators. The first three U.K. licenses were initially awarded to Unitel, Microtel, and Mercury. After some reorganization, some commercial shake-outs, and some arrangements that split up coverage areas, only two operators, Mercury One-2-One (formerly Mercury and Unitel) and Hutchison Orange (formerly Microtel), now offer PCN services based on DCS 1800 technology. Investment logic and intensive market research led to this step in the direction of common sense after it was discovered that the PCN cake was not big enough for three consortia laying out three complete networks competing against two digital cellular operators (Cellnet and Vodafone) and two analog cellular operators (Cellnet and Vodafone). All of this occurred shortly after telepoint system’s (using CT-2 cordless technology) disastrous failure in the United Kingdom. After 2 years of aggressive network rollout, and even more aggressive marketing, both PCN operators have together already earned a cellular market share of 17% in early 1997. This figure looks
58
GSM and Personal Communications Handbook
even better when we compare the PCN share of close to 42% among only the digital GSM-based standards (GSM and DCS), which says that most new customers are attracted by the PCNs. Most of the first subscribers were found in metropolitan areas (cities and major towns) where network rollouts began in order to satisfy the mandate of the licenses that required a certain degree of coverage within a relatively short time. In other countries, some PCN licensees are also GSM operators, and certain coverage obligations imposed by regulations may be achieved through dual-band services, that is, GSM for wide-area coverage and DCS for busy spots, office campuses, and in-building service. This will be an attractive and low-cost alternative for solving the GSM 900 operators’ coverage problems once dual-band GSM 900/DCS 1800 phones become available. GSM operators can use the spectrum in the DCS 1800 band to overcome capacity constraints in their GSM 900 networks. Sufficient quantities of dual-band GSM 900/DCS 1800 phones need to be available before this scheme can work; a significant number of users must be driven off the 900-MHz band, or most of the relatively small number of heavy users must be forced onto the 1800-MHz band. It is the European Union’s policy that member countries must license at least one DCS 1800 operator by January 1, 1998, either to new network operators or to existing GSM operators. Roaming is a major issue for PCN operators. Indeed, the DCS 1800 operators need the 900-MHz roaming resources to alleviate their capacity problems and fill their coverage holes more than the 900-MHz operators covet the DCS 1800 band. DCS 1800 customers can already enjoy SIM card roaming when they use their SIMs in rented GSM 900 phones abroad, but actual station roaming awaits dual-band terminals. For now, a GSM subscription is more competitive in international roaming situations. One of the reasons the PCN operators are so strong in the United Kingdom is that they tried to distinguish their service from that of the GSM operators from the first day. The differentiation went beyond different marketing schemes, positioning of the service, and pricing; they offered somewhat different features and services from the 900-MHz operators. The PCN operators were very active in setting their own service standards and adding even more functionality to the GSM standard. The availability of a second line and number (alternate line service) for each subscriber (one for private use and another for business use) is just one
From Pan-European mobile telephone to GSM
59
example. Besides their own association (European PCN operators) the PCN operators are also represented in the GSM MoU Association through their own committee called the Personal Communications Networks Interest Group (PCNIG). 2.1.5
PCS 1900
The PCN experience in North America was, and remains, about as different from the European one as anyone can imagine. The reader should remember that GSM is a living standard that includes specifications for an entire network; it is not just an air interface definition. We are still modifying and enhancing the specifications to meet new expectations and markets. Turning his gaze from the orderly GSM world in Europe to the relative chaos of the American wireless landscape, the casual observer might throw up his arms in confusion over the collection of sundry incompatible air interfaces. The European and American approaches have their own peculiar logic with global consequences. Whereas the Americans allow different radio systems and technologies to fight each other in the market place, the Europeans confine the competition to the standards process. GSM was thoroughly tested in the early 1990s within the confines of the wealthy European community with the result that GSM was presented to the rest of the world as a fully functional, low-cost, low-risk way to deploy modern digital cellular service. The disadvantage of the orderly European approach is that it takes time; GSM does not necessarily represent the latest and greatest in digital wireless technology. The American way is to curtail sharply the lengthy standards process, and allow system proposals to prove themselves in the cold reality of the market. This fast prototype approach furnishes a stage on which the newest technologies can mature, but moves the expense of testing and proving those new technologies closer to the operators and their customers. So, since most of the world is not buried in mountains of excess cash that can fund elaborate experiments in radio technology, we see GSM deployed in more than 100 countries. The more high-tech American proposals, which are reviewed in Chapter 3, can only survive in more robust economies or under the auspices of a concerted national effort. Why, then, did GSM appear in North America? Spectrum was finally reserved in North America for PCSs in the 1900-MHz band in 1994. The 1800-MHz band was too clogged with, for example, point-to-point government radio users to consider this more international frequency
60
GSM and Personal Communications Handbook
allocation. Chapter 1 explained the novel licensing approach for PCSs in the United States and the complex matrix of frequency allocations based on MTAs and BTAs. The frequency bands themselves are shown in Figure 2.1. We see that two 60-MHz bands—Base RX and Base TX, separated with an 80-MHz duplex offset—are dolled out in pairs (RX and TX) of 15- or 5-MHz allocations lettered A through F. Low-powered (inbuilding) PCS operation is also allowed in the unlicensed band, which is seen in Figure 2.1 as a 20-MHz resource between the licensed downlink and uplink allocations. Operators are free to deploy any kind of radio system they can afford as long as the selected technology does not interfere with other licensees within the PCS bands or with radio services outside those bands. Each system has its own way of defining a physical channel within the American PCS spectrum (some of these schemes are disclosed in Chapter 3). The selected technology can even be a TDD technology that does not use separate uplink and downlink bands. The selected radio technology must, however, be disciplined enough to guarantee users a useful service. Unworkable and inefficient systems cannot continue to hold a license. The original field of almost a dozen proposals was quickly reduced by economic reality to three major contenders: IS-95 (CDMA), IS-136 (TDMA), and PCS 1900 (GSM). The place of the IS-95 and IS-136 contenders, together with a few other interesting systems that may see new life in the future, is explored in Chapter 3, where we will see that the AMPS system retains an important influence on all PCS technologies and operators. The PCS 1900 system is based on GSM Phase 2 specifications; even the channel numbering scheme was adopted from DCS 1800. In PCS 80 MHz Base RX A
1850
D
B
E F
Base TX C
unlicensed
1930
1865 1870
1885
1910
A
D
E F
C
1945 1950
1890 1895
Figure 2.1
B
North American PCS frequency bands.
1965 1970 1975
1990
From Pan-European mobile telephone to GSM
61
1900, the absolute radio frequency channel number (ARFCN), which is N in the following formulas, takes on the following values: 512 07:15 AF 1551 10:00 AF 1501 11:00 LH 4360 Select Confirm
Confirmation 14-03-98 MUC-CDG 07:15 AF 1551 Boarding-No. 105 Seat 1A C-Class
(a)
(b)
Figure 9.2
9.1.6
(a) Flight selection and (b) electronic boarding pass.
Conclusion
In this section on the SIM application toolkit we demonstrated that this GSM feature offers an abundance of new services and applications to the GSM subscriber that are particularly attractive to users. It is an easy exercise for the operators to define and introduce new services with the SIM application toolkit. Differentiation of network services among operators is supported. The advantage for the operator is that she can define and implement features independently from the mobile phones’ manufacturers. Once the manufacturers support the SIM application toolkit the operators can use off-the-shelf products to realize their valueadded services. Specific applications might ask for such differentiation in different countries and for different user groups. The SIM application toolkit is a box of tools with which different combinations of different features can be constructed to offer many new services and benefits.
9.2 Customized applications for mobile network enhanced logic (CAMEL) In increasingly competitive GSM market environments, operators want to develop their own new value-added services, so-called operator-specific
358
GSM and Personal Communications Handbook
services (OSS). But OSS only work satisfactorily if customers have access to them even when roaming. Because subscribers get used to special features, they will not understand why networks have different implementations or even different legal requirements such as for calling line identification. Customized applications for mobile network enhanced logic or CAMEL was defined to support OSS while roaming. CAMEL lets a GSM network offer IN features. This is made possible through the exchange of relevant information between, for example, the HLR and VLR. The logic providing the services and the call processing functions are separated from each other, which makes provisioning new services easier since the switch-related signaling functions remain transparent. CAMEL-related standards will be introduced through GSM Phase 2+ in two steps (see Chapter 4). 9.2.1 Functional description of CAMEL
CAMEL makes dedicated network features available to roaming subscribers even though a visited PLMN may not support them. A popular example of this is dialing into the voice mail. Most networks nowadays use a short network internal number (e.g., “3311” mentioned in Chapter 6) rather than a complete national or international number (e.g., +49511XXXXXXX) through which the subscriber gets voice mail. The only problem with this approach arises with roaming subscribers, since these numbers may not exist in the visiting network or they may have another significance. CAMEL would extend the reach of the home network’s unique dialing sequences to roaming customers. How can this work? The CAMEL service environment (CSE) controls all outgoing and incoming calls to a CAMEL subscriber regardless of her location. Whenever a roaming customer registers in a foreign PLMN, the HPLMN provides the visited PLMN with relevant information on the subscriber, most of which should be familiar to us: authentication parameters, call forwarding parameters, etc. The CAMEL subscription information (CSI) is added to the CSE information for a CAMEL subscriber. The CSI notifies the visited network about specific features that have to be supported for the visiting subscriber. CSI contains an identifier for OSS that is transparent to the visited network. CAMEL applies to all circuit-switched basic services except for emergency calls, which can be placed by anyone in any network.
New Phase 2+ functions
359
The HPLMN has to be notified for each call setup with the CAMEL subscriber regardless of whether it is a mobile-originated (MO) or mobileterminated (MT) call. When the VPLMN realizes a call setup involves a CAMEL subscriber, it suspends the call and informs the HPLMN (with the CSE) about the event, the parties involved in the setup, and the location of the calling subscriber. The CSE at the subscriber’s HPLMN provides the VPLMN with information on how to proceed with the call: to continue with the call unchanged, to bar the call, or to continue with modified information (e.g., a different called party as indicated in our example on the voice mail). All this has to happen with minimum delay. Together with some operator-specific services, the CSE must investigate the whereabouts of a subscriber from time to time. The HPLMN will inform the CSE about the current subscriber status (idle, busy, detached) and the location information. As with all advanced features, so it will also be with CAMEL that some networks provide certain services before others do, but CAMEL also requires that some signaling links dedicated to CAMEL be supported even if nothing else is implemented. When the HPLMN realizes that a CAMEL subscriber wants to register in a network that does not support it, the HPLMN has the ability to apply operatordetermined barring or deny location updating if necessary. Which option is finally exercised depends on the CAMEL subscriber’s subscription profile. The CAMEL subscriber will, in most cases, at least be able to use his phone without his OSS in the visited network. The specifications for CAMEL are detailed in [2,3]. 9.2.2
Network architecture
Now we look at how CAMEL is implemented in the home (HPLMN) and visiting (VPLMN) networks. Figure 9.3 depicts the different entities and functions that are required. The main signaling link between the HPLMN and the VPLMN is through the two GMSCs of both networks. The home location register (HLR) stores the CSI, including both mobile-originated and mobile-terminated calls, for each of the individual customers who subscribe to the HPLMN’s service. The HLR provides these data to the visitor location register (VLR) when a mobile station starts a registration procedure or when the CSI is updated. The GMSC will also get the data when routing information is required during a call setup to or from the mobile station. The GSM service control function (gsmSCF) is a support function for CAMEL allocated to the HLR. It keeps track of the current CAMEL status
360
GSM and Personal Communications Handbook
gsmSCF
HLR
GMSC
HPLMN
VPLMN
VLR
Figure 9.3
GMSC gsmSSF
Network architecture required for CAMEL support.
of a subscriber and tracks the location data. The GSM service switching function (gsmSSF) is an interface inside the GMSC that detects requirements for CAMEL support, for example, a roaming CAMEL subscriber initiates an outgoing call. The VPLMN’s VLR gets a copy of the CAMEL data needed for outgoing calls. 9.2.3
A CAMEL example
We now illustrate the uses of CAMEL with a trivial example. Many network operators add service numbers to the SIM cards they supply to their subscribers. Some examples of service numbers are (1) the voice mail number, (2) the network’s operator, (3) the network’s help desk or customer care center, (4) a travel service, (5) a pizza delivery service, and (6) a flower delivery service. The service numbers do not follow the numbering rules for standard phone numbers; they may be four-digit numbers that only have significance within the home network. If a customer wants to use one of the service numbers stored on the SIM while roaming, he will not be able to get through to his home system because the meanings of the service numbers are totally unknown to the visiting operator. When CAMEL is applied to the situation the call control is located in the home country, which knows all about each of the service numbers; the
New Phase 2+ functions
361
roaming subscriber is properly connected to the service he seeks, which would probably be the travel service rather than the pizza delivery service in most roaming examples. This simple example should clarify how CAMEL works. The same mechanism applies to all operator-specific features that will be offered in the future.
9.3
Railway applications
The features and applications described in this section are not limited to railway use even though they meet the specific needs of the Union Internationale de Chemin de Fer (UIC). The chief requirements of the European railways dealt with priorities and preemption for call setups, the speed of call setup procedures, and simple procedures for group call services and voice broadcast services (see Chapter 3). These requirements ensure that GSM can respond to their internal demands for a safe and secure railway communications system based on a European-wide standard that specifies the same communications gear in all countries. Even though UIC features will be available in future GSM-based systems, we should note that the railway organizations will implement their networks themselves on their own frequencies set aside for the purpose, and will not simply add something to an existing GSM network. An additional 4 MHz of spectrum below the extended GSM frequency band is reserved for the railways. 9.3.1 Enhanced multilevel precedence and preemption
The enhanced multilevel precedence and preemption service (eMLPP) actually consists of two services: The precedence service gives certain calls priority, and the preemption service makes sure that calls of high priority get the required resources. The eMLPP service belongs to the group of supplementary services to which users subscribe as they need them. The service is applicable to speech and fax teleservices as well as to all bearer services [4]. 9.3.1.1
Priority levels and preemption
Seven priority levels are defined as indicated in Table 9.2. Levels A and B are used only within the network and we cannot subscribe to them. Both
362
GSM and Personal Communications Handbook Table 9.2
Priority Levels Defined for eMLPP Priority Level
Purpose
A
Highest, for network internal use
B
For network internal use
0
For subscription
1
For subscription
2
For subscription
3
For subscription
4
For subscription
the A and the B levels are mapped to level 0 outside of the network area. The other five levels are used not only within the network but also apply globally, which means they are also applicable in an ISDN, which provides the multilevel precedence and preemption service (MLPP); see [5–7] for further details. The maximum priority level for an individual subscriber is stored in the SIM. The user may select the level she wants to use for a specific call or she may set a default value for all calls. The user cannot select a priority that exceeds the maximum level specified by the priority level. The advantage for a user with high priority is that when the system is congested or there are limited radio resources in a part of the network, sufficient resources are freed up so that the priority user can proceed with his call. The resources are taken away from users without an eMLPP subscription or from eMLPP subscribers with a lower priority level. Lower priority users who have their traffic channel resources taken away from them simply hear a congestion tone. The decision on which resource should be preempted is made by the MSC. The reason for introducing such a service is that certain customers, such as a railway, must ensure that service or a traffic channel is available whenever needed. An engine driver cannot wait to report a problem; human lives or the general public safety may be in jeopardy. 9.3.1.2
Call setup time improvements
A logical extension of the priority requirement is the demand for improved call setup times. Normal call setup times range from a few
New Phase 2+ functions
363
seconds up to 15 or 20 seconds, which is the time between the instant the send button is pressed and the instant a ringing indication is heard from the other side of the connection; this may be too long in critical or emergency situations. Additional classes for call setup times have been specified as listed in Table 9.3, which reflects the allocation for UIC requirements. The classes are allocated to the priority level by the operator. It should be obvious that call setup times of less than 1 second are very difficult to achieve in GSM. These times will be achieved for 95% of all calls in case of railway networks; the remaining 5% of the calls may suffer setup times 1.5 times longer. The setup times for class 1 calls depend on the required service. Broadcast or group calls, which are described later, demand setup times of less than 2 seconds. These fast setup times are achieved by omitting authentication and ciphering procedures. Figure 9.4 shows the message exchange requirements for normal GSM calls and Figure 9.5 shows the procedures for high-priority calls. The reader will note the diminished number of transactions in Table 9.4 relative to those in Table 9.3. Because messages are not only transmitted over the air interface but have their origins somewhere in the network, it is obvious that fewer messages reduces the time required to set up a call. 9.3.1.3
Automatic answer for eMLPP
The eMLPP service allows a mobile station to automatically answer incoming calls of certain levels. The parameter for the level of calls to be answered thus is stored in the SIM. If the called user is busy on the phone, the eMLPP service can preempt the current call and put it on hold in order to connect a call with a higher priority. A subscription to eMLPP should always include a subscription to call waiting and call hold. The only call that does not allow preemption is an emergency call.
Table 9.3
Call Setup Performance Class
Setup Time
Priority Level
1
< 1 sec
A
2
< 5 sec
B, 0
3
< 10 sec
1, 2, 3, 4
364
GSM and Personal Communications Handbook
MS channel request
network channel assignment
service request authentication request authentication response ciphering mode command ciphering mode complete set-up call proceeding TCH assignment TCH assignment complete alerting connect
connect acknowledge communication
Figure 9.4
Call setup for normal calls.
MS channel request
network channel assignment
service request set-up
TCH assignment TCH assignment complete communication
Figure 9.5
9.3.1.4
Call setup for high-priority calls.
Railway (UIC) applications
To summarize the eMLPP service and the setup times allocated to the respective levels, we will examine some typical UIC applications, the details of which can be found in [8]. Table 9.4 not only gives examples, but also shows that the preemption applies only to the first four priority levels, which, looking at the examples, should get very high priorities. The allocations given in these example apply to UIC requirements; it is the operator’s choice to allocate the setup time class or allow preemption for individual priority levels. This kind of service is useful for any large group that has dedicated tasks in our society and will make use of the good coverage of GSM networks.
New Phase 2+ functions
365 Table 9.4
Examples of eMLPP Service Priority Level
Setup Time
Preemption
UIC Example
A
Class 1
Yes
Railway emergency application, broadcast/group call or even a data call (e.g., to stop a train)
B
Class 2
Yes
Automatic train control
0
Class 2
Yes
Emergency services
1
Class 3
Yes
Operational train calls
2
Class 3
No
Train data services
3
Class 3
No
Railway information services
4
Class 3
No
Passenger services (e.g., phone booth in train)
9.3.2
Voice group call service
GSM already offers the multiparty supplementary service that mimics a kind of voice group call service. The voice group call service (VGCS) introduced by UIC reflects their different requirements. The building of a multiparty communications link takes much more time than can be tolerated in an emergency and there is a limit to the number of participants. VGCS allows call establishment times of a few seconds to all members of a group and offers half-duplex service to limit the network resources used for the service [9,10]. 9.3.2.1
Service description
The members of a voice group are fixed by the network operator or the service provider upon the request of the subscriber. Members of a group are identified by the group identification or group ID. Individual subscribers can be members of several groups. Where and how the group ID is stored in the network is detailed in the following section. VGCS distinguishes different participants of a group call. A dispatcher is not necessarily a mobile subscriber but can also be located in the fixed network; he will always be addressed when a group call is placed to his group. The dispatcher has several privileges that are discussed later. The
366
GSM and Personal Communications Handbook
calling subscriber is the one who initiates a group call; individual subscribers need to have authorization to initiate group calls. The destination subscriber is anyone who receives a group call. A voice group call can be initiated by a dispatcher or by any authorized subscriber. The call setup is performed via a normal traffic channel. The members of the group are notified (paged) via a dedicated notification channel (NCH). When all the members of the group are connected, or after a predefined time if some members are missing, the initiator of the group call will be advised that the group call is established. All members of the group are now listening to a dedicated voice group call channel on the downlink. This channel is the same for all members within a cell, which drastically reduces the required resources. After the initiator has finished talking he will lose the dedicated link and join with the others in the voice group channel. From now on only one member can talk on the uplink. An uplink channel is allocated upon request. When several members start talking at the same time, the network will select one on a first come, first served basis. The voice group call service is a half-duplex teleservice. VGCS also offers security against eavesdropping. In this case the uplink and the downlink channel apply ciphering with a group key that is identical for each member of the group. Dispatchers are the only exception to use of the voice group call channel; they always have a full-duplex channel and can, therefore, always talk. The network will forward the dispatcher’s voice without regard to a group member who may already be talking. If there are several dispatchers connected to a group all of them can talk and all of them will be heard by the group members at the same time. This is possible because such calls are linked within switches (MSCs) with one dedicated link for each of the individual members. Figure 9.6 illustrates the voice group channel and the associated half-duplex links. In our example a central location (the switch tower in a railway yard) starts a voice group call. The two trains in our example are paged and connected to a voice group channel on which they would be able to listen to the message coming from the switch tower. When the announcement is finished the dedicated link to the switch tower is terminated and replaced by the same half-duplex resource allocated to the trains. Now everybody gets an uplink channel upon request from which individual acknowledgments of the message are received from the trains. The only exceptions are the dispatchers who might talk at any time.
New Phase 2+ functions
Train = MS
367
Network
Train = MS
Fixed line Switch tower = Initiating MS
Figure 9.6
9.3.2.2
Dispatcher
Voice group call.
Network architecture
The only addition to the standard GSM network for the support of VGCS is the group call register (GCR). The GCR is logically collocated in the MSC. The interface between the GCR and MSC has not been standardized and depends on individual manufacturer implementations. The GCR may, therefore, be an integrated part of a MSC. The GCR stores all data related to the VGCS (see Figure 9.7), such as the following: ◗ A list of cells that belong to the service area. For UIC applications it is
obvious that only cells along the train tracks will belong to the service area. ◗ A list of relay MSCs required to cover the complete service area. ◗ A list of members for individual group IDs. ◗ Parameters for group calls, for example, a time-out value when calls
will be terminated after no activity, information on the cipher algorithm, and the group key to be used for a group call.
368
GSM and Personal Communications Handbook
cell area
BTS
HLR BSC
ISDN Network (PSTN)
BTS GMSC
BSC BTS
MSC
GCR
Figure 9.7
VLR
BTS
VGCS infrastructure.
Members are only identified via the group ID; the temporary mobile subscriber identity (TMSI) or international mobile subscriber identity (IMSI) have no meaning. The group ID is used to page members of the group. The group ID is broadcast for the duration of the voice group call to allow members entering the service area to join the group call. 9.3.3
Voice broadcast service
The voice broadcast service (VBS) is very similar to the VGCS except that VBS only allows announcements from one party to the remaining members of the group; there are no responses from the group. The network entities involved in VBS are identical to those used for VGCS. The GCR contains a separate set of parameters identifying the VBS. The SIM contains separate data to identify different options for VBS. Further details can be found in [11,12].
New Phase 2+ functions
369
References [1] GSM 11.14, “European Digital Cellular Telecommunications System (Phase 2+); Specification of the SIM Application Toolkit for the Subscriber Identity Module–Mobile Equipment (SIM-ME) Interface,” ETSI, Sophia Antipolis. [2] GSM 02.78, “Digital Cellular Telecommunications System (Phase 2+); Customized Applications for Mobile Network Enhanced Logic (CAMEL); Service Definition (Stage 1),” ETSI, Sophia Antipolis. [3] GSM 03.78, “Digital Cellular Telecommunications System (Phase 2+); Customized Applications for Mobile Network Enhanced Logic (CAMEL) —Stage 2,” ETSI, Sophia Antipolis. [4] GSM 02.67, “Digital Cellular Telecommunications System (Phase 2+); Enhanced MultiLevel Precedence and PreEmption Service (eMLPP) —Stage 1,” ETSI, Sophia Antipolis. [5] ITU-T Recommendation I.255.3, “ISDN MultiLevel Precedence and PreEmption (MLPP) Stage 1.” [6] ITU-T Recommendation Q.85, “Stage 2 Description for Community of Interest Supplementary Services (Clause 3: Multi-Level Precedence and Pre-Emption MLPP).” [7] ITU-T Recommendation Q.735, “Stage 3 Description for Community of Interest Supplementary Services Using SS No. 7 (Clause 3: Multi-Level Precedence and Pre-Emption (MLPP).” [8] GSM 03.67, “Digital Cellular Telecommunications System (Phase 2+); Enhanced MultiLevel Precedence and PreEmption Service (eMLPP) —Stage 2,” ETSI, Sophia Antipolis. [9] GSM 02.68, “Digital Cellular Telecommunication System (Phase 2+); Voice Group Call Service (VGCS)—Stage 1,” ETSI, Sophia Antipolis. [10] GSM 03.68, “Digital Cellular Telecommunications System (Phase 2+); Voice Group Call Service (VGCS)—Stage 2,” ETSI, Sophia Antipolis. [11] GSM 02.69, “Digital Cellular Telecommunication System (Phase 2+); Voice Broadcast Service (VBS)—Stage 1,” ETSI, Sophia Antipolis. [12] GSM 03.69, “Digital Cellular Telecommunications System (Phase 2+); Voice Broadcast Service (VBS)—Stage 2,” ETSI, Sophia Antipolis.
CHAPTER
10 Contents 10.1 Routing in GSM PLMNS
Roaming and call routing
Roaming and call routing
10.2 Charging principles 10.3 Phase 2+: support of optimal routing (SOR) 10.4
Conclusion
G
SM is a mature system accepted throughout the world. Most of us who travel with our GSM phones take it for granted that when we cross borders, or switch on our phones after leaving the plane, we have reliable access to mobile telephone service. With all the GSM roaming agreements in place and many manufacturers making dual- and triple-band phones, service coverage will no longer be limited to Europe but will extend to Asia, the Pacific Islands, Africa, the Middle East, and even to North America. It does not seem to matter where we switch on our GSM telephone—we are connected and people can find us; we get service and we can make calls. Today’s reality sounded like dreams or the ravings of unbalanced people just 10 years ago. Some of us have grown to depend on the global nature of GSM, and others in the industry base their wealth and their ability to meet their monthly rent payments on the miracle. How does all of this work? What are the mechanisms behind roaming? 371
372
GSM and Personal Communications Handbook
In this chapter we explore how calls are routed to our mobile phones with seemingly little regard for our location, and we briefly consider how we are billed for this convenience. Because such costs cannot be expressed in absolute figures, we will give an indication of which part of a call is billed to the calling party and which part is billed to the called party.
10.1 Routing in GSM PLMNs Because mobile users are, by definition, always on the move, the mechanisms the networks use to locate each of us within the GSM world are important. We will confine our attention to how calls are forwarded to our mobile stations, which are also referred to as mobile-terminated (MT) call setups. We will check different cases as we see how the networks route calls to specific mobile stations [1]. 10.1.1
Location registration
The most important tool used for finding phones within GSM networks is location updating. The location of a mobile station is uniquely identified by the mobile country code (MCC), the mobile network code (MNC), and the location area identity (LAI). The MCC is a three-digit value that identifies the country where the network is located. The MNC is a two-digit value (three digits in North American PCS 1900 systems) that identifies different (competing) networks within one country. The LAI identifies the physical area in which a mobile station is located. A location area may consist of one or more physical cells; it is also referred to as a paging area, which a network designates as an MS’s location for paging tasks. With each location updating procedure, the mobile station reads its location from the control channel transmitted by its serving base transceiver station (BTS) and reports it back to the network. In a GSM network, as shown in Figure 10.1, two registers store the location-related data of a mobile station: the home location register (HLR) and the visitor location register (VLR). The HLR keeps data that are permanently associated with individual mobile stations as well as the current location. The permanent data include subscription details such as the teleservices, bearer services, and supplementary services allocated to the subscriber. The billing information and the subscriber’s home address are not in the HLR; they
Roaming and call routing
373
GSM PLMN Location Area 2
HLR BSC
BTS
VLR
GMSC
BTS
BSC
ISDN / PSTN
Location Area 1
BTS
MSC VLR BSC BTS BTS
Figure 10.1
Location Area 3
Network infrastructure of a GSM network.
are located elsewhere in the billing center. The VLR keeps temporary data on a subscriber for only as long as the subscriber is located in the area belonging to a particular VLR. The temporary data contains the subscription-related data (obtained from the subscriber’s HLR) as well as the MS’s exact location in the VLR’s area. Figure 10.1 shows two VLR areas. Location area 1 and location area 2 belong to the first VLR, and location area 3 belongs to the second VLR. Note that the resolution of the location areas is the same as the base station controllers (BSCs). As an aid to understanding the hierarchy of locations (MCC, MNC, and LAI), we briefly explain the situation in North America’s PCS 1900 systems by noting the large size of the North American continent and the likely possibility that a single North American PCS 1900 operator may have multiple HLRs. In the United States (MCC = 310), there are several operators, two of which are VoiceStream [2] and Omnipoint [3]. VoiceStream operates systems all over the continental United States including one system thousands of kilometers away on the Hawaiian Islands. It
374
GSM and Personal Communications Handbook
seems logical, therefore, that VoiceStream maintains several MNCs (200, 210, 220, 230, 240, 250, and 260). Omnipoint runs a large system in the New York City region (MNC = 160) and operates a much smaller system in Wichita, Kansas. The subscription details for the Wichita subscribers resides in the HLR located in New York such that all of Omnipoint’s subscribers, as of this writing, have the same MNC. A network may consist of several mobile services switching centers (MSCs), but only those connected to an ISDN or a PSTN are used as gateway MSCs (GMSC), with an attached HLR, since this is the switch through which incoming calls enter the system. The incoming calls for mobiles in the network cannot be routed properly until the system’s HLR is checked for the location of the target mobile and the subscriber’s authorized services. The GMSC and the MSC each have their own VLRs. Upon performing a location update, the message from the mobile station that marks its location is first sent through the MSC to its attached VLR. The VLR determines if it already has a record of the MS and, if so, only the location area is changed if it has, in fact, changed from an earlier one. If the MS is not known to the VLR (it is a new visitor to the system), then the VLR will apprise the mobile’s HLR of the MS’s location and routing information as it sends along a request for the subscriber’s home data, which the visited VLR will need before it can complete a mobile originated call from the mobile. After all of this is done, the HLR knows which VLR should receive incoming calls for the roaming mobile and the VLR knows the exact location of the mobile and which BTS should send paging messages that attract the mobile’s attention. If, later, the mobile wants to originate a call, the VLR has a record of the subscriber’s subscription data, that is, authentication variables and service options. There are three different types of location updating procedures [4]:
1. The registration takes place whenever a mobile station is switched on. After an internal initialization, including necessary SIM procedures, the MS checks for an available network. When it finds one the MS is able to read important information such as the location information. Two cases are distinguished here: (1) IMSI attach required or (2) IMSI attach not required. In the first case, the IMSI attach required case, the mobile station always initiates a location update procedure. A location update procedure is also completed in the second case (the not required case) but only when the
Roaming and call routing
375
location area identity the MS reads from the downlink control channel has changed since it was last switched off. In the first case, the mobile also starts an IMSI detach procedure when it is switched off, thus telling the network that the MS is no longer idle, that it cannot respond to incoming calls. This prevents the squandering of radio resources when mobiles that cannot possibly respond are paged. In the not required case, the detach procedure is not required. The first case should be the rule; the second case the exception.
2. Periodic location updating is performed after a period of time predefined by the network and constantly sent to all active mobile stations monitoring the control channel. If the mobile station does not register after this time, perhaps because the subscriber drove into an underground parking lot, then the network will assume that the mobile station is no longer available and will mark it as “not reachable” in the HLR and VLR. This means that incoming calls for this MS will not be routed to the location area but will be blocked at the home GMSC. If the subscriber invoked call forwarding on subscriber not reachable, the network will forward the call to the indicated number. 3. When the MS detects a location area change it will notify the network that it is now located in a different area. As shown in Figure 10.1, an MS could move around within location area 1, from one cell to another, without the need for a location update. If it moves from location area 1 to location area 2 the VLR must be notified. If the MS moves from location area 1 or 2 into location area 3, then the HLR must be notified of the change and the VLR in area 3 will store the mobile’s new LAI. 10.1.2
Routing within a PLMN
Figure 10.2 shows the routing of a call from mobile subscriber A to mobile subscriber B. When A calls B’s number the call will be routed to the GMSC of the network (1). If the network has more than one GMSC, then the call will be routed to the GMSC to which subscriber B belongs, that is, to the GMSC the attached HLR of which hold’s subscriber B’s permanent data (2). The distinction between the HLRs and the attached GMSCs is made through the phone number—the first digit(s) after the network access
376
GSM and Personal Communications Handbook
GSM PLMN VLR
HLR
2
3
5
GMSC
A
Figure 10.2
1
6
MSC
4
BSC
BTS
7
B
Routing of a call within a GSM PLMN.
code. The GMSC will discover the location of subscriber B as it determines which HLR responds to its location request. The responding HLR tells the inquiring GMSC which MSC area or location area the subscriber is currently visiting (3). With this information the GMSC can route the incoming call to the particular MSC that can complete a proper call to subscriber B (4). Since the MSC area is covered by many BSCs and BTSs, it has to check (5) the exact location of B within the VLR, which passes the latest subscriber information, including the location, back to the MSC (6). With this information the MSC is able to forward the call to subscriber B (7) through the proper BTS. Routing of a call from the fixed network follows the same principle. The only difference is that the GMSC will be accessed from an ISDN or a PSTN. 10.1.3 Call routing when a mobile station is roaming
When the mobile station is roaming, the process includes a foreign network to which calls have to be properly routed. Figure 10.3 shows an example of a call from the fixed network to a roaming mobile subscriber. The originator of the call does not usually know where the called mobile is actually located and may not even know the called number is a mobile phone that could be in a different country roaming in a foreign network. The ISDN/PSTN subscriber A sets up a call to GSM subscriber B by dialing his normal PLMN access number. The call is routed through the
Roaming and call routing
377
country 1
country 2
GSM PLMN GSM PLMN
ISDN / PSTN HLR
3
LEC
A
1
Figure 10.3
VLR
4 6
GMSC
2
7
GMSC
5
BSC
BTS
8
B
Routing a call to a roaming subscriber.
local exchange center (LEC) to the HPLMN (1 and 2). The GMSC asks its HLR where the subscriber is currently located (3 and 4) and routes the call to the (G)MSC that serviced B’s last known location (5). After determining the exact location (6 and 7), the call is eventually routed to subscriber B (8). The interesting and important fact is, as stated earlier, that A will not notice that he is actually calling into a foreign network. A special case applies when B (normally in a country foreign to A) is roaming in the home country of A, which is depicted in Figure 10.4. This time A calls B, which is an international call because B is permanently registered in a foreign country. The call is routed to B’s home PLMN. From there the call is routed back to the originating country because B’s HLR knows that B is actually located there. Here we have call routing with two international links.
10.2
Charging principles
The standardization body for GSM agreed on charging principles on which networks should base their charges to their customers [5]. The home network operator will collect all charges an individual subscriber incurs, including those from foreign networks. A general rule is that the calling party pays. This does not only hold true for a GSM subscriber, but
378
GSM and Personal Communications Handbook
country 2
country 1
country 2
GSM PLMN GSM PLMN
ISDN / PSTN HLR
3
LEC
A
1
4 6
GMSC
2
Figure 10.4
VLR
7
GMSC
5
BSC
8
BTS
B
Special case for routing a call to a roaming subscriber.
also applies to callers in the fixed network calling into a GSM PLMN. The rate that is fixed for calling out of or into a GSM PLMN should cover the investment for the infrastructure that allows the call to be completed in the first place. Thus calling a mobile subscriber from a fixed phone is usually more expensive than completing a long-distance call to a fixed-line phone from the same fixed phone. But calls made from a mobile phone are usually more expensive. Just like all rules in this world there are exceptions to general practice; rules are made to be broken. The exceptions are discussed in the following sections as we make distinctions according to the location of a mobile subscriber and the mechanisms the networks use (e.g., call forwarding) to connect callers to mobile subscribers. In the examples that follow we always assume a fixed-line subscriber A, and a mobile subscriber B. 10.2.1
National call charges
When both parties, the calling and the called party, are located in the same country, the calling party always pays the charges incurred. Figure 10.5 distinguishes both cases; when subscriber A calls subscriber B and the other way around. Both subscribers get their respective monthly bills from their individual operators. In some countries the same operator
Roaming and call routing
379
GSM PLMN ISDN / PSTN
LEC
A
GMSC
B A originates and pays B originates and pays
Figure 10.5
National call charges.
might run both the mobile and the fixed networks so that he might receive a combined bill. 10.2.2 Call charges when roaming
When the GSM subscriber, B, is roaming he also has to pick up the additional international part of the costs, the so-called roaming leg. The roaming leg is the part from the home PLMN to the visiting PLMN. The reason for allocating the call charges to the called party is that there is no way the caller could be aware of B’s location in another country. Subscriber A still has to pay for the part from the ISDN/PSTN network to the GSM PLMN. The situation in which subscriber B calls subscriber A is handled as a normal international call where B has to pay the international rate of the visiting country plus a roaming surcharge, which will be collected by his home operator as a reward for properly handling the roaming customer. Figure 10.6 illustrates the situation. The worst case for charging occurs when subscriber A tries to call roaming subscriber B, who lives and pays his monthly GSM bill abroad, but is currently located in A’s home country, as shown in Figure 10.4. In this case A incurs international call charges (A should know where B lives) and B has to pay the international leg back to A’s country. A cheaper connection appears when B calls A. Subscriber B has to pay only a local call charge for the noninternational call through the local GSM network plus a roaming surcharge to his home operator.
380
GSM and Personal Communications Handbook
Country A GSM HPLMN
ISDN/ PSTN LEC
Country B GSM VPLMN
GMSC
GMSC
A
B A originates and pays
B pays the international part
B originates and pays the international call
Figure 10.6
10.2.3
Call charges when roaming.
Call forwarding
The third case applies when subscriber B activates the call forwarding supplementary service that was explored in Chapter 7. When A tries to call B he will be charged for the call into the GSM network. The remaining part from the GSM network to the forwarded-to-number will be charged to subscriber B. 10.2.4 More exceptions to the rule
The three examples just given should serve only as guidance. There are, of course, many more cases where a subscriber incurs charges that are at variance with the “caller always pays” rule. We are reminded that a monthly subscription fee is collected from subscribers by mobile network operators that covers the expenses associated with making the network available day and night, making sure that monthly bills are accurate, and paying for improvements to the network that make it even more valuable. That there are also exceptions to the rule becomes more evident with the example of India where the mobile subscriber always pays [6]. It is considered normal practice among a growing number of PCS 1900 operators in North America to charge nothing for the first minute of incoming calls to mobiles. Subscribers are spared the expense of answering “wrong numbers” or unwanted calls, and are thus not discouraged from giving
Roaming and call routing
381
out their numbers, which increases traffic. Many of us turn to complaining at bill paying time; we complain that the charges for mobile phone services are too expensive. It is at such times that we can consider the alternative: how would we have to alter our daily lives if the mobile network we have grown to depend on were to suddenly and completely fail or disappear. What is its value to us then?
10.3 Phase 2+: support of optimal routing (SOR) In previous sections we saw that call routing does not always follow the most direct route, particularly in the case of roaming subscribers (see Section 10.2.2). In addition, the call charges for nonoptimal routing are quite high, and the traffic unnecessarily occupies and ties up international lines. The international link is paid twice in some cases. There must be a better way. One small example should help as an introduction to this section about optimal routing. Two employees (A and B) of the same company are on a business trip in a foreign country, say, Italy. Both A and B have their mobile phones with them and both are registered in their home country, say, Germany. If A, who is attending a meeting in a hotel in Rome, wants to communicate with B, who is finishing a report at another hotel in Rome, A has to call B through his home number (in Germany), which results in an expensive international call, and B has to pay, again, for the incoming roaming leg from A (through Germany). This is a rather expensive call if both are located in the same city! GSM Phase 2+ brings a solution to this kind of situation that will be implemented through support of optimal routing (SOR). The intention of SOR is to reduce unnecessary international links. The initial phase of SOR will focus on two situations:
1. Optimal routing of calls within one country, which applies to the extreme but not uncommon example presented here; 2. Optimal routing of calls to the country where the call normally would have been routed, and which especially applies to conditional call forwarding.
382
GSM and Personal Communications Handbook
Optimal routing applies to all teleservices and bearer services, but not to dedicated PAD and dedicated packet access. Callers from fixed networks will not benefit from optimal routing since their networks do not have the ability to investigate a GSM PLMN and to apply optimal routing. Only GSM mobile subscribers, who typically pick up lots of extra charges, will benefit from SOR. If one of the networks involved in a connection does not support SOR the normal GSM call routing will apply. The advantage of SOR is reduced unnecessary international call traffic and a resulting reduction in call charges. Each call record for optimal routed (OR) calls or call forwarding will contain the connection charge (for the reduced distance) and an OR flag indicating that optimal routing was used. Further information is found in [7,8]. The impact of optimal routing is illustrated by some examples where we use mobile subscriber A as the calling party, mobile subscriber B as the called party, and subscriber C as the forwarded-to-party or voice mail server, which could be a mobile or fixed subscriber line. 10.3.1 Roaming mobile subscriber
A and B are mobile subscribers from different countries. A calls B, who happens to be located in A’s home country. Under normal circumstances both subscribers would have to bear the high international call charges associated with these kinds of calls. With optimal routing applied, A’s HPLMN would try to route the call to B’s HPLMN, but B’s HPLMN reports back to A’s HPLMN that B is currently located in country A. So, A’s HPLMN withdraws the call and routes it directly to subscriber B (in country A) rather than through country B, as shown in Figure 10.7. If A is a fixed-line subscriber rather than a mobile phone subscriber, SOR would not be applied and both would pay the higher call charges. 10.3.2 Call forwarding to home country
B is currently traveling in a foreign country C, and B has activated conditional call forwarding to his voice mail located in his home country. Another mobile subscriber, A tries to call B. The call is routed to the visited country C, where a call forwarding condition is met. Because the call is
Roaming and call routing
383
A
country B
B optimal routed link interrogation
country A
standard GSM routing
Figure 10.7
Roaming subscriber B located in A’s home country.
redirected back to the home PLMN, optimal routing applies and B covers only the local call charges. This situation is depicted in Figure 10.8.
country B
A B country A
mail box country C
optimal routed link interrogation standard GSM routing
Figure 10.8
Roaming subscriber B with conditional call forwarding.
384
GSM and Personal Communications Handbook
10.3.3 Call forwarding to visited country
Subscriber B, who lives in country B, is currently visiting a subsidiary of his company in a foreign country A, and has activated conditional call forwarding to a secretary C, located in the visited country (A). Subscriber A, who lives in country A, tries to call B (in country B). The call is routed back to the visited country (country A) where a call forwarding condition is met and the call is forwarded to C (Figure 10.9). Because the call is redirected to the home PLMN of subscriber A, optimal routing applies directly to secretary C, and subscriber A has to cover only local call charges.
10.4
Conclusion
Here ends our look into the networks, at call routing, at some call charging issues, and at SOR. In the next and last chapter we pull the mobile phone from the subscriber’s hands and, over the subscriber’s persistent objections, remove the phone’s battery and pry off the covers to discover how it works and begin to see how we can make one for ourselves.
country A
A
country B
B
C
optimal routed link interrogation standard GSM routing
Figure 10.9 Roaming subscriber located in the same country as the calling subscriber with call forwarding to a local number.
Roaming and call routing
385
References [1]
GSM 03.04, “European Digital Cellular Telecommunication System (Phase 2); Signaling Requirements Relating to Routing of Calls to Mobile Subscribers,” ETSI, Sophia Antipolis.
[2]
http://www.voicestream.com.
[3]
http://www.omnipoint.com.
[4]
GSM 03.12, “Digital Cellular Telecommunication System; Location Registration Procedures,” ETSI, Sophia Antipolis.
[5]
GSM 02.20, “European Digital Cellular Telecommunication System (Phase 1); Collection Charges,” ETSI, Sophia Antipolis.
[6]
Shankar, N. K., “Regional Focus: India,” Mobile Communication International, Issue 33, July/August 1996, p. 17 ff.
[7]
GSM 2.79, “Digital Cellular Telecommunication System (Phase 2+); Support of Optimal Routing (SOR); Service Definition (Stage 1),” ETSI, Sophia Antipolis.
[8]
GSM 3.79, “Digital Cellular Telecommunication System (Phase 2+); Support of Optimal Routing (SOR); Technical Realization,” ETSI, Sophia Antipolis.
PART
III GSM technology and implementation
CHAPTER
11 Contents 11.1 Breaking GSM down 11.2 Transmitters and receivers 11.3 MS and BTS— new roads to the ultimate radio 11.4 Baseband signal processing 11.5 Speech coding and speech quality in GSM 11.6
Equalizers
11.7 Encryption and security in GSM 11.8
Mixed signals
11.9 Microprocessor control 11.10
GSM timing
11.11 Components and technology 11.12 Guide to the literature
Introduction to GSM technology and implementation
Introduction to GSM technology and implementation
E
ven though the TDMA concept was not new when GSM was introduced to the world along with some other secondgeneration “digital” wireless systems, the industry had to take some big steps into new territories. The first halting and plodding steps were in the direction of getting everything to work. A mountain of details had to be sorted out: the radio techniques had to be implemented, speech codecs needed to be tested, and systems had to be designed in the light of actual commercial practice. The steps became a brisk walk as the industry transformed the technology so that it could be reproduced under different manufacturing circumstances, and it broke into a run as it slashed the cost of the technology so that millions of people all over the world could afford it.
389
390
GSM and Personal Communications Handbook
The different implementations of digital mobile telephones made tremendous progress in only a few years. Just compare the size, weight, and cost of the early models of only a few years ago with those available today. A glance into the trade publications proves that progress is far from over. The network components, base stations, switches, location registers, antennas, and supporting systems have sustained equal transformations even as they are hidden from the gaze of those who buy digital phones. No single chapter in any book can do justice to the fascinating technology and extraordinary accomplishments of the industry, but we can show the way. What are the issues we should consider when we design a modern digital cellular product? What are we getting into? All GSM phones have specific jobs to do. What are their functions, and how do we implement them? What are the typical building blocks that can make up the functions, and how do they work? There are about as many ways to build a digital cellular phone as there are people who endeavor to do so. Different implementations are carried out through different combinations of methods, technologies, and architectures; each approach has its own advantages. Rather than give advice on how to design certain functions, we treat the salient aspects of the technology, and point out valuable references. Though we explore the whole GSM system, we concentrate on the phone itself: the most visible part of GSM, which, as far as the subscriber is concerned, is the GSM system. We offer detailed explanations of important and uniquely digital functions in GSM phones that are often left to specialized literature. One of the functions is the speech codec, which, though it is well known and the object of many informal discussions and marketing hype, is surrounded by myths. Another one is the forgotten and mysterious equalizer, which, since it uses so many important digital radio techniques, we cover in some detail. With these two “hard parts” exposed and explained, the remainder of the phone is reduced to (1) variations on the functions and techniques used in the speech codec and equalizer and (2) traditional radio functions and circuits (mixers, amplifiers, oscillators, etc.). The aim is not to offer up all the details required to build a GSM phone, but to achieve a general understanding of the building blocks, architectures, and implementations; to point the way into the technology rather than explain it all Benchmark figures are given where necessary. Only a basic
Introduction to GSM technology and implementation
391
understanding of communications theory and related system matters is expected of the reader. The last section of this chapter is a guide to the references and further reading.
11.1
Breaking GSM down
Unlike some introductory books that consider only the broad system aspects of GSM, just as this book does, we reserve some space in this final chapter to probe into the building blocks, functions, devices, and techniques of digital radio. 11.1.1 Physical and logical blocks of a GSM mobile station
As in any communications systems, a digital cellular system’s purpose is to provide reliable, cost-effective transmission and reception of the user’s information. Digital cellular systems carry user information, including speech, computer or message data, and even facsimile information in digital form. Even though we talk broadly about a digital communications system, the physical transmission from a digital radio transmitter to a digital receiver is analog, and so is the audio interface to the users on both sides of the channel. The digital processes are confined within the phone. If we look inside a GSM MS we can identify a number of building blocks that process audio signals and user information before transmitting it. Some other blocks work in the opposite direction because they convert the received signals back into the form in which they entered the transmitter at the other end of the link. A measure of a phone’s performance can be how “invisible” the phone appears to the user: the voice from the receiver’s speaker should sound exactly as it did at the transmitter’s microphone, and computer data should arrive at the receiver free of errors. Even as we strive to make the phone invisible, we need some kind of control and interface functions to do so. Figure 11.1 shows a logical block diagram of a GSM mobile station. This figure concentrates on the physical and logical signal processing flow. The blocks in Figure 11.1 represent functions, not necessarily actual devices. As we will see, because many of the signal processing functions share similar processes and techniques, only a few physical devices perform the tasks.
392
GSM and Personal Communications Handbook
VAD
SPEECH ENCODER
ADC
COMFORT NOISE FUNCTION
SPEECH DECODER
DAC
EXTRAPOLATION Voice Codec
Speech Processing
Figure 11.1 processing.
11.1.1.1
Class 1 CRC Conv. Coding Re-ordering Interleaving Cyphering Burst Build. GMSKModulation
TRANSMIT PA MXR
DAC
Filter RECEIVE
Equalizer Burst Disass.
ADC
De-Cipher De-Interleave Decoding Class 1 CRC Signal Processing
LNA MXR Filter
Baseband Codec
Radio Section
Block diagram of a GSM mobile station signal
The voice codec
The voice codec performs the conversions between the radio’s speech processing functions, which are digital processes, and the analog audio domain outside the phone. We find two converters in the voice codec: (1) the analog-to-digital converter (ADC) for the analog input signal from microphone and (2) the digital-to-analog converter (DAC) for the recovered analog output signal to speaker. The output of the voice codec’s transmit path is the digital input to the GSM speech codec. The input to the speech codec is a 13-bit representation (13-bit resolution) of the voice signal sampled at 8 kHz. This 13-bit linear pulse code modulation (PCM) is represented by a 104-Kbps data stream (13 × 8,000 = 104,000), which means there are 8,000 samples per second, where each sample is represented by 13 bits. The voice codec contains band-limiting low-pass filters for antialiasing in the transmit path, and waveform reconstruction in the receive path. There are also some gain amplifiers for the signals coming from the microphone and the signals that drive the speaker. Multiplexers can be
Introduction to GSM technology and implementation
393
used in order to switch acoustic signal paths, for example, from a built-in handset microphone or speaker to a hands-free microphone or speaker. 11.1.1.2
GSM speech processing
The GSM full-rate (FR) speech codec (speech encoder and decoder) compresses the transmit side’s 104-Kbps voice signal into only 13 Kbps by drastically reducing the amount of redundant information from the voice codec [1,2]. The 13-Kbps speech data rate is more suited for transmission over the limited radio resource, which is a sharp contrast to the typical ISDN (fixed-line) PCM signal, which represents speech at a relatively high rate of 64 Kbps. For GSM speech encoding/decoding on the fixed network side, a conversion between the fixed-line 64-Kbps (8-kHz sampling rate × 8-bit resolution) and the GSM 104-Kbps speech codec (8-kHz × 13-bit resolution) is done by companding and expanding the digital representation of the audio in accordance with the A-law or µ-law scales. A compander converts the higher resolution 13-bit values to 8-bit values by quantizing low-amplitude signals more precisely than high-amplitude values. Lowamplitude samples are given a higher resolution with smaller step sizes, and high-amplitude values get a lower resolution with larger step sizes. The reverse process of compressing is called expanding, hence, companding. Further significant reductions in redundant speech information and data rates call for a much more intricate processes beyond the simple quantization process; the final reduction in payload data occurs in the speech coder. The counterparts of the mobile terminal’s speech codec can be found in the networks in different physical places, for example, with the BTS, BSC, or MSC, although it is logically associated with the BSS. The placement depends on network topology and line traffic considerations. The transport of speech data on the network takes place either in encoded logical 16-Kbps subchannels (13 Kbps + synchronization data, submultiplexed from 64-Kbps channels), which is the most economical way in terms of line traffic, or in standard 64-Kbps channels. The GSM full-rate speech coder is a type of hybrid coder (see Section 11.5.4) called a regular pulse excited long-term prediction codec (RPE-LTP). The output of the GSM speech coder occurs in frames of 260 bits at a rate of one every 20 ms (260 bits/20 ms = 13,000 bps). This
394
GSM and Personal Communications Handbook
speech data (separated into class I and class II bits) is passed on to the channel coding process, which prepares the coded speech bits for transmission on the radio channel [2,3]. The reverse side of the process in the receiver sees 260-bit frames coming from the channel decoder, which are used by the speech decoder to recreate the speech sounds that specified the bits the coder generated. Auxiliary functions such as voice activity detection (VAD), extrapolation, and a comfort noise function support features such as discontinuous transmission (DTX) and substitutions for lost frames. DTX is an effective way to save battery capacity by minimizing transmit cycles when no voice activity is detected. This feature also reduces interference in adjacent cells. Two more speech coders are specified for use in GSM networks. One is the half-rate (HR) coder for use in half-rate channels [4]. GSM half-rate frees up the use of every other time slot in a single full-rate traffic channel, thus doubling the channel capacity. The enhanced full-rate (EFR) coder uses the same rate as normal full rate [5]. It was required by the North American personal communications industry for use in PCS 1900 networks. The EFR algorithm that was chosen (known as the US-1 codec), and also adopted by ETSI (through GSM Phase 2), offers much better perceived voice quality. General speech coding techniques, and the GSM speech coders in particular, are explained in considerable detail in Section 11.5. 11.1.1.3
Signal processing
The signal processing function works with the user data (speech or data traffic), as well as signaling and control information. Payload data must be well protected before transmission over the radio channel, and the protective measures must be “undone” on the received data. The GSM channel coding block protects speech coder information with several processes: cyclic redundancy check (CRC) for the most important class I bits, followed by convolutional coding, and interleaving. User data (other than voice data) are transmitted over GSM data channels, which are convolutionally coded and interleaved with their own unique schemes, depending on the data rate, that are different from those used on voice data. Signaling data are block coded, convolutionally coded, and block interleaved according to their own rules. A common ciphering scheme can be applied to the channel coded data. The channel decoding block reverses the transmitter’s coding for the received data. The burst building and multiplexing process
Introduction to GSM technology and implementation
395
adds a midamble (training sequence code), tail bits, and guard bits to the encoded bits after the interleaving, and makes sure that the burst data (radio bits) are delivered to the modem and the radio at the right time. The radio bits are differentially encoded and modulated through a Gaussian filter function as part of the modem function [6]. The reverse side of the modem function is dominated by the channel equalizer, which adaptively compensates for the adverse effects (intersymbol interference) applied to the original signal through time-variant multipath interference, delay spread, and Doppler shifts. This function is explained in Section 11.6. After deciphering and deinterleaving the data from the equalizer, a decoding algorithm recovers the original payload data. 11.1.1.4
Baseband codec
The baseband codec converts the transmit and receive data into analog or digital signals, respectively. The transmit process delivers analog baseband I- and Q-signals (from two DACs), which modulate the radio carrier [2]. The receive process converts the analog I- and Q-signals from the radio back to a digital data stream, which is filtered, sampled, and quantized (again, two ADCs) before presentation to the equalizer. A typical value for the converter’s resolution is 10 bits for both directions. 11.1.1.5
The radio section
The radio section, which includes the IF stages, is one of the interfaces to the outside world; the other is the audio interface. Functions such as frequency synthesis, a local oscillator, up-and-down conversion with mixers (MXR in Figure 11.1), analog RF and IF filtering, transmitter power amplification (PA in the transmit path of Figure 11.1), preamplification in the receive path (LNA in Figure 11.1), RF pulse shaping, automatic gain control, and frequency correction and control are handled in the radio section. A general view of the radio section is offered in Section 11.2 of this chapter. 11.1.1.6
Other functions
Apart from routine signal routing and conditioning tasks between audio, baseband, and radio blocks, there are some other functions that complete the GSM mobile station. Some control units are required to organize and schedule logical and physical entities. On top of the physical layer
396
GSM and Personal Communications Handbook
functions, which should be viewed as the physical conditioning and processing of payload data (coding and decoding), we need to accommodate some higher level tasks such as (1) synchronization, (2) frequency and time acquisition, and tracking, (3) monitoring (serving cell and adjacent cells), (4) received signal strength measurements, and (5) radio control. Above these radio tasks we need to add the higher protocol layers (layers 2 and 3) with signaling functions, which allow the phone to operate in an orderly way within a network. Other control functions within a GSM mobile station handle the user interfaces (keypad, display, and the user menu), the SIM interface, and other auxiliary (data) interfaces. 11.1.2 Physical and logical blocks of a GSM base station
A GSM base station is the mobile’s counterpart, but not in all its aspects. The base station does not provide for most of the layer 3 signaling functions, and none of the higher layer functions, because it is transparent to those functions coming from the network. Because the base station converts the network’s wire environment to an air interface, base stations exercise their own control over the layer 1 and layer 2 processes. The signal processing functions in a GSM base station are similar to those in a mobile station. Additional logical channel structures, not found in the mobile station’s uplink, must be supported by a base station: (1) common control channels (CCCH), which comprise the paging channel (PCH) and the access grant channel (AGCH); (2) the broadcast control channel (BCCH); (3) the frequency correction channel (FCCH); and (4) the synchronization channel (SCH). All of these are unidirectional channels; a mobile does not support these channels beyond being able to read and understand them. The interface to the fixed network requires some wireline physical and logical entities that provide transmission and reception capabilities for the user data, signaling data, and control information. One of these entities is the transcoder rate adaptation unit (TRAU), which supports the multiplexing of speech data to and from the Abis interface [7]. Just as there are many configurations possible within a GSM network, so there are many architectures of GSM base stations. There are different applications, different numbers of channels, different configurations for standby channels, and different network interface configurations. Figure 11.2 shows only one example, which assumes that the speech codec is placed remotely in the network (e.g., within the mobile
Introduction to GSM technology and implementation
397
TRANSMIT PA MXR Filter
DAC
Class 1 CRC Conv. Coding Re-ordering Interleaving Cyphering Burst Build. GMSKModulation
TRAU Abis Interface
RECEIVE LNA
ADC
MXR Filter Radio Section
Figure 11.2 path.
Baseband Codec
Equalizer Burst Disass. De-Cipher De-Interleave Decoding Class 1 CRC
TRAU
Abis
Abis Interface
Signal Processing
Sample block diagram of a GSM base station signal
switching center), and shows the architecture for a single transceiver. The appropriate signal processing and radio sections would exist in parallel in multiple-transceiver architectures.
11.2 Transmitters and receivers The radio sections in cellular phones and base stations require an enormous design effort. Circuits need to meet specifications even as they are manufactured in large quantities at low cost. In particular, this means that analog circuitry must be produced with an absolute minimum of individual “tweaking” (tuning and calibration) exercises such that millions of “equal” radios are the result. Radio design, which is strictly separated from the digital design task, is sometimes regarded as “black magic” or “alchemy” rather than a disciplined and orderly engineering task. These
398
GSM and Personal Communications Handbook
kinds of references arise from myths surrounding RF work that are given prolonged life as “lab legend” by those unfamiliar with the techniques and skills of RF practice. Radio design is a balancing act in which the successful practitioners find the optimum combination of technology, technique, cost, and performance. The skill and experience of the attending engineer is the key to meeting design goals. Figure 11.3 shows one sample block diagram of a digital radio structure that may be used in a GSM transceiver. Here we find a single conversion transmitter and a dual conversion, superheterodyne receiver with quadrature (I/Q) modulation and demodulation, respectively. Many other architectures and techniques are possible, and the reader is referred to existing literature on the subject. Section 11.12 is a starting point for those new to radio work. The most important aspects, parameters, functions, components, and requirements are discussed in the following sections. 11.2.1
Transmitters
There are two kinds of radio transmitters in GSM: (1) the ones in the BTS, which create the physical forward side of the RF link to the mobile station (downlink), and (2) those we find in the MS, which create the other—the
BB-TX-I
Modulator, IF Mixer & Amplifier
BB-TX-Q
IF IF Band Pass Filter
IF
RF Up RF RF Band VGA ConPass +PA verter Filter
TX Spurious Filter
ANT. RF
IF LO VCO
RF LO VCO
PLL(s)
PWR. CTRL.
Antenna Switch or Duplexer
Reference Oscillator IF2BP Filter BB-RX-I BB-RX-Q
IF2
I/Q DeModulator
Figure 11.3
IF Mixer & AGC
IF1 IF1BP Filter
BandRF IF1 RF DownPass Converter Image Reject
RF RF
LNA
RX PreSelector Filter
AGC
A GSM radio transceiver architecture example.
Introduction to GSM technology and implementation
399
reverse—side of the RF link back to the base site (uplink). The chief difference between these are (1) the much greater RF power, relative to the MS, generated in the BTS, and (2) the tiny size of the MS with the proximity of the transmitter to the MS’s receiver. The similarities outnumber the differences. Those who design transmitters for GSM service are obliged to consider the following parameters:
1. The power output is seldom more than 2W for an MS, but it can be more than an order of magnitude greater in a BTS. The amplification job usually needs to be distributed over several stages in the high-power applications in base stations [8]. The transmitter’s output power needs to be managed over temperature as well as with changes in the source impedance of the power source for the RF amplifier stages, and the RF output components (including the antenna). 2. Power consumption (efficiency of amplifiers) is critical in the MS. 3. Load pull is the transmitter’s tendency to change its output frequency with impedance changes at the antenna, or in the power source of the transceiver [9]. 4. Modulation accuracy (linearity) is compromised by attempts to increase the efficiency of the transmitter [10]. 5. SNR, hum, and noise are general terms for those factors that limit the best signal-to-noise ratio (SNR) that we can expect from a transmitter. Linearity problems in the modulation conspire with noise sources and unwanted RF coupling to put a limit on our best efforts. 6. Spurious outputs can be either harmonic ones or all of those undesired outputs that are not harmonics of the desired output [9]. The former usually come from linearity problems in the RF amplification stages of the transmitter, and the latter find their origin in mixers and frequency synthesizers. 7. Excessive adjacent channel power comes from too much single sideband (SSB) phase noise from the transmitter’s oscillator, or poor fractional frequency stability performance of the frequency synthesizer [9].
400
GSM and Personal Communications Handbook
8. Intermodulation distortion appears when a strong signal from outside the transceiver is coupled into the transmitter, through the antenna, and mixing products are created in one of the many nonlinear mixing sites anywhere in the transmitter’s signal path. This is a particularly troublesome problem in base sites, because base stations often share site space with other high-powered transmitters such as TV transmitters. Seven of the transmitter’s functional components that are not a part of the digital baseband functions are listed next in the order in which they appear as the transmit path approaches the antenna:
1. An oscillator is one of the reference inputs to a phase-locked loop (PLL) synthesizer. 2. The frequency synthesizer functions as an intermediate/radiofrequency voltage-controlled oscillator (IF/RF VCO), which is a means to add frequency agility to the fixed reference oscillator. The relatively high RF levels of the transmitter’s output reduce the required noise performance of the synthesizer compared to what is required of the same function in the receiver’s local oscillator (LO). 3. The modulator imparts intelligence to the carrier. GSM’s GMSK modulation affords many possibilities for modulators, but some kind of I/Q process is the general method [10]. I/Q modulators are sometimes referred to as “universal modulators.” We could, for example, use a more direct technique than an I/Q method to generate MSK, such as a pair of oscillators operating 67 kHz on either side of the assigned channel frequency, and then use the modulator’s data stream input to enable one or the other oscillator. Chief among the many reasons such direct methods are not used in digital radio is that the frequency shifts are sloppy and abrupt. The more complicated I/Q method can be made to generate a wide variety of digital modulation schemes including GMSK. They work by enlisting two double-balanced mixers into which we apply alternate halves of the symbol data stream; first into one mixer, and the next into the other mixer. Each mixer has a reference input of the transmitter’s carrier frequency shifting from
Introduction to GSM technology and implementation
401
each other by 90 degrees. The outputs are summed and passed into the up-converter. The utility of this quadrature technique is that the phase and the amplitude of both parts of the data input can be adjusted and filtered through look-up tables before they are applied to the mixers.
4. An up-converter (mixer) brings the transmitter’s output up to the assigned frequency. There are many kinds of mixers, and all of them produce odd order products that have to be filtered from the transmitter’s output [10]. The double balanced variety is often used, because it isolates the frequency synthesizer from the variable load of the RF power amplifier. 5. Amplification raises the transmitter’s output to the required RF level specified for the task. The job is made particularly difficult in the GSM mobile station, because the power has to be adjusted over a wide range and intermittently keyed within strict burst power-time templates without the AM splash that usually accompanies intermittent duty transmitters. All of this has to be accomplished in a small volume with a limited power source (battery). 6. Filters attenuate most kinds of spurious outputs before they reach the transmitter’s antenna, but they have none of the effects of unwanted outputs caused by linearity problems in the modulator. Filter design has become a very specialized skill, and the possibilities are enormous [8]. 7. Though the half-duplex operating mode of GSM may seem to eliminate the need for a duplexer, it is often included in mobile stations. The duplexer is a filter that is common to the transmitter and the receiver, and it is included to (1) add receiver front-end filtering to reduce the harmful influences of nearby transmitters in other phones and (2) increase the isolation between the transmitter and the receiver. (See the same subject in Section 11.2.2.) The transmitter’s performance characteristics, which we listed in this section, are mostly a matter of regulatory compliance. Designs must comply with the requirements, but there is no compelling reason to increase costs by improving on them. When our transmitter designs fail to meet
402
GSM and Personal Communications Handbook
the required minimum performance we cannot expect type approval. The requirements are, however, not capricious and arbitrary barriers designed merely to increase the costs of radios offered into GSM service; they are designed to make the best use of the spectrum, and yield a balance between system capacity, costs, and the best quality of service the technology can deliver. 11.2.2
Receivers
The wide use of traditional radio technology and techniques in the RF functions contrasted with the use of integrated functions on silicon for the baseband functions that we examined in the transmitter is also evident in receivers. The receiver’s job is to extract the signal from a distant transmitter and recover the modulation. Once the modulation is recovered, advanced techniques (1) equalize and decode the channel, (2) separate signaling from traffic information, (3) recreate the voice sounds that entered the transmitter, and (4) bring audio or user data out of the transceiver. GSM’s minimum requirement is to be able to recover a transmitter’s symbol stream with less than only 1 bit error in 100 for signals of less than –102 dBm, which is only 6.3 × 10–14 W. The noise present at the input of a GSM receiver will be, at best, about –121 dBm. This bit error ratio (BER) requirement as a criteria for receiver sensitivity is rather modest when compared to wireline systems where one error in one million symbols is considered intolerable. Because channels in wireline systems are relatively stable and orderly compared to those in wireless systems, the protection coding in wireline systems would be too weak to be able to respond to the error rates typical in radio channels. Current GSM receivers perform far better than the minimum –102-dBm requirement. Typical implementations achieve the required BER performance at nominal levels of between –105 and –110 dBm. Outside system influences, including the presence of strong blocking signals, or co-channel and adjacent channel interferers, Doppler and deep fades, quickly reduce margins in the designs. To recover information with low BER from a signal that is suffering interference from a co-channel signal (from a distant base station or mobile station using the same physical channel), the available signal bandwidth should be high. The opposite is the case for adjacent channel interference, which typically occurs at higher levels than co-channel interference. Here we want to have an as narrow as possible filter
Introduction to GSM technology and implementation
403
characteristics in the signal recovery path in order to separate the good signal from the bad ones. (See Figure 11.5 in the following pages for the effect of IF filters that perform this task.) Surface acoustic wave (SAW) filters are the device of choice for IF filters. They are piezoelectric transducers on which interfering mechanical waves are set up such that the devices become bandpass filters with sharp cutoff characteristics. SAW devices have high insertion losses (about 20 to 30 dB). The most narrow bandwidth characteristic is selected such that it can still pass the received signal without distorting the symbols that eventually appear from the demodulator; too narrow a filter will hurt the BER performance of the receiver. In general, improved system performance in wireless networks is limited to optimizing receivers, not the transmitters, and all kinds of tricks and techniques are brought to receiver designs to improve BER performance. The improvements tend to be confined to the baseband processes (DSPs or ASICs) after the demodulator or detector. All air interface standards go through a maturing process, such that we can see, as a rough rule, a steady improvement in the average BER performance of all the receivers deployed in a particular kind of air interface technology of about 1 dB each year. Eventually, we see receiver improvements slow down such that each 1-dB improvement takes much longer than a year to achieve, and each improvement quickly becomes more and more costly. This slowing in receiver performance improvements occurs as they approach fundamental thermal noise limits. When we finally observe this slowing in receiver improvements, then we declare the host air interface technology to be “mature,” and we can start to look for spurts of innovation elsewhere in the systems that wring additional capacity, performance, and quality from the system by further lowering average BERs throughout the systems. We also see increased interest in new air interface proposals as system improvements slow and become more expensive. A receiver fulfills its duties by processing the small amounts of energy at the antenna such that the original information is recovered with minimum distortion and errors. The specifications that are used to judge a receiver’s particular abilities to do its work are discussed in the following paragraphs. The sensitivity of a receiver refers to its ability to react properly to a weak signal. Digital receivers use the maximum BER at some low RF level
404
GSM and Personal Communications Handbook
as a measure of their performance. This is analogous to the SNR technique used in analog receivers [9,11]. The SNR technique is not useful in digital receivers because they have some kind of decision circuit early in their signal recovery and analysis stages that makes constant determinations on what the actual transmitted symbols probably were. If the BER = 1%, then the decision circuit makes one bad decision for each 100 symbols it is called on to judge (see Section 11.6). The selectivity is a receiver’s ability to reject unwanted signals that are very close to the frequency of the desired signal [9]. These interfering signals are usually spurious emissions from nearby transmitters. The duplexer and the IF stage’s SAW filter are the devices of interest here. The average price of phones can be held down with strict adherence to outside system matters such as low transmitter adjacent channel power emissions, strict frequency planning, and proper system design. Intermodulation rejection is the receiver’s ability to overcome its own natural tendency to generate an internal on-channel signal from offchannel signals present at the antenna. The receiver’s spur-free dynamic range is a measure of its immunity to intermodulation, which is primarily determined by the receiver’s input stages. Figure 11.4 illustrates the derivation of the IP3, or third-order intercept point, and the spur-free dynamic range of a receiver. A receiver’s nonlinearity can be measured by injecting two closely spaced signals (f1 and f2) of increasing but equal amplitude into the RF input (antenna port) while observing the rise in third-order intermodulation products at both 2(f1) – f2 and 2(f2) – f1. Intermodulation products appear in the receiver when the input test signals exceed whatever level drives the receiver beyond its spur-free dynamic range. Figure 11.4 shows that the level of the third-order intermodulation products (the dark curve with the greatest slope in the figure) rises at a rate three times greater than the rise in the input signals (the dark curve to the left of the steep curve). We confine our attention to the third-order products because all of the other intermodulation products are far outside the receiver’s own passband. As the amplitude of the input signals increases, the level of the receiver’s output, both the fundamental signals and intermodulation products, increases until the input stages start to saturate (the curved parts of the dark traces in the figure). If we extrapolate the straight parts of both curves in Figure 11.4 beyond the saturation levels, then the point at which both lines intercept is called the third-order intercept (IP3).
Introduction to GSM technology and implementation
405
output level (dBm)
spur-free dynamic range
input level (dBm) IP3
Figure 11.4
Receiver dynamic range.
The further to the right IP3 is in Figure 11.4, the better the receiver’s intermodulation rejection performance. Receivers exhibiting poor intermodulation performance will loose sensitivity; we say the receivers suffer from desensitization, as they pass close to transmitters in a different network from that whose services the subscriber is using. Image rejection is a measure of the receiver’s ability to attenuate signals that appear in the receiver’s IF stages, or output, as a result of an offchannel signal with the same offset from the LO frequency as the desired input signal. Figure 11.5 illustrates how images are rejected from a receiver’s output. The top part of the figure depicts a spectrum analyzer display on which we see our desired signal (made fatter then the other signals for ease of identification) together with others that are not particularly interesting to us and need to be filtered out. The frequency of our LO is plotted in the center of the scale. The second part of the figure has the passband of the RF preselection filter, our duplexer, plotted on the scale; signals
406
GSM and Personal Communications Handbook Desired Signal
RF Spectrum LO
0 Hz
IF
Desired Signal RF Preselector
RF Spectrum LO
0 Hz
IF Desired Signal IF Filter
IF Spectrum − IF
0 Hz
+ IF
Desired Signal
IF Spectrum − IF
Figure 11.5
0 Hz
+ IF
Image rejection.
outside the passband are attenuated. The third part of the figure plots the downconverted output of the receiver’s IF where the LO frequency translates itself to 0 Hz. The subsequent IF filter’s passband is plotted on the right side of the “IF spectrum” scale together with another image channel
Introduction to GSM technology and implementation
407
the mixer translates to –IF. The bottom part of Figure 11.5 depicts the result out of the IF stages. Since the receiver’s detector cannot distinguish between the desired signal and the image, it accepts the superposition of both signals. If the lower side of the preselector filter is less then 2 × IF away from the desired signal (the condition shown in the second part of the figure), then the image will be attenuated. Raising the IF will allow more freedom in preselector specifications but will demand IF filters with a smaller fractional passband specification. With the exception of the demodulator or the detector, the devices and functions in the receiver have the analogous duties they had in the transmitter, but in the reverse order. The chief difference is that since the signal levels are millions of times smaller than they are in the transmitter, the required performance of the amplifiers, mixers, and oscillators (including the synthesizer) is higher. Just as we did in the transmitter, we will follow the signal from the antenna back toward the user [9,10]:
1. A duplexer is a three-port filter. The antenna is attached to a common port, and the receiver is attached to another port where we find low loss at the receiver’s operating frequencies between the receiver port and the antenna. The transmitter is connected to the remaining port where another low-loss path, at the transmitter’s operating frequencies, is found to the antenna. The filter is constructed in such a way as to isolate the transmitter from the receiver. Even though the TDMA/TDD timing in a GSM mobile station calls for a minimum idle time of two time slots between transmit and receive instances, the isolation the duplexer affords between the transmitter and the receiver is valuable. 2. A preselector filter limits the bandwidth of the receiver, which reduces intermodulation (IM) distortion. Additional preselector filtering is sometimes required to reduce the cost of the duplexer. 3. The RF amplifier—also referred to as a low-noise amplifier (LNA)— determines the IP3 of the receiver, and isolates the preselector filter from the image filter that follows the amplifier. As we discovered in Figure 11.4, the location of IP3 is a way we can judge a receiver’s ability to limit any harmful effects off-channel signals may have on the receiver’s performance. Intermodulation distortion is generated in active components in the receiver’s input
408
GSM and Personal Communications Handbook
circuits, which act as mixers for the input signal and another offchannel signal that yields an output within the receiver’s pass band. The RF amplifier raises the level of the input signal together with the noise. Improvements in intermodulation performance of any RF amplifier are usually accomplished at the expense of additional current requirements for the amplifier, thus reducing battery capacity. The influence of the noise a device can contribute in a circuit also fights our efforts to optimize a receiver’s treatment of weak signals. Space restrains us from including a treatment of these important influences, but the reader is invited to refer to [9] and also Chapter 4 in the same reference for guidance.
4. An image filter can follow the RF amplifier to attenuate spurious signals and images. It also attenuates harmonic distortion generated in the RF amplifier. 5. The LO is the mix-down reference for the input signal, and it is generated in the same frequency synthesizer that creates the transmitter’s carrier, but not on the same frequency. An injection filter between the LO and the mixer (see next item) will attenuate noise from the LO. 6. The mixer accepts the LO’s input together with the amplified and filtered input signal to generate a lower frequency copy of the input signal. The output is called the intermediate frequency (IF). There are many kinds of mixers, each with their own advantages and disadvantages [9]. 7. The IF amplifier and appropriate control logic are responsible for most of the analog gain of the receiver, which is constrained with an automatic gain control (AGC) function. 8. The detector or demodulator is the last analog function in the receiver. It reverses the transmitter’s modulation process in order to recover the original I- and Q-signals, which are sent to the baseband processes as shown in the lower left corner of Figure 11.3. Designing a receiver is an elaborate balancing act that starts with allocating gains, losses, and signal levels over all eight blocks in the receiver’s
Introduction to GSM technology and implementation
409
RF and IF paths. The process is reduced to an orderly one in [9], which is highly recommended for the reader new to this kind of work. As the design proceeds, the designer will find herself clearing spurious signals generated in the last mixer stages, which starts the balancing act over again. Interest is growing in using digital techniques in front of the baseband chips in receivers. One of these new techniques is the digital IF, which replaces the last mixer in a typical analog superheterodyne receiver with an analog-to-digital (A/D) converter. If the first IF is, say, 10.7 MHz and we are interested in designing a wideband receiver that can be used for a variety of services in addition to cellular and PCS ones, then the A/D converter must perform 21.4 million samples per second. If we limit the IF bandwidth, as we would if we are designing an alternative handset for PCS services, we can reduce the sampling rate. All further processing duplicates the analog mixer, but performs it in the digital domain. The I and Q baseband components can be extracted from the digitized 10.7-MHz IF by performing all the tasks an actual analog I/Q demodulator accomplishes within a virtual digital copy of the process. Such a virtual machine requires digitized sine and cosine versions of the reference signal from an oscillator or frequency synthesizer, and can recover the I and Q components by performing multiplying operations between the digitized IF and sine/cosine references. Digital filters can notch-out unwanted signals, and then, once the result is decimated in proportion to the reduction in IF bandwidth, DSP processes can be applied as they are today with current techniques. Readers new to digital radio are usually surprised by the wide use of old and traditional radio technology and components used in the RF sections of base stations and phones [12]. This practice has two origins. The first is that optimizing any of the circuits and functions listed above is not a particularly straightforward and orderly task: impedances change for a variety of reasons, circuits tend to oscillate even when they are not designed to do so, and ground loops and shielding problems are cleared through experimentation. Once the RF functions finally work, they tend to work very well. It is often easier and cheaper to “borrow” an old circuit from another radio than to design a new one. Second, most of the components in the RF parts of phones are passive devices that are very small, cost very little, and have excellent quality. There is currently no cost advantage in employing integrated RF functions.
410
GSM and Personal Communications Handbook
As the power consumption of a portable phone becomes an increasingly important differentiating factor in the market, transmitter performance is examined more closely. Standby times in mobile phones are limited by the receiver and the baseband section’s performance. Exploiting functions like discontinuous receive (DRX), in which the receive path in an idle state needs to be activated only intermittently (for reading paging channels or monitoring serving and adjacent cells), yields long standby times. Talk time is greatly influenced by the efficiency of the RF transmitter. Taking advantage of the fact that average speech activity, which can be detected with voice activity detection (VAD), is typically around 45% during a normal speech conversation helps save battery power. Enlisting VAD in GSM is called discontinuous transmission (DTX). We are still left with the power amplifier’s efficiency (loss vs. actually radiated power) as the remaining parameter to improve as best we can. The efficiency of a transmitter’s RF amplifier remains the chief limiting factor in battery capacity today; but when DSPs were first employed in radios, they contributed a significant contribution to battery drain. The efficiency of RF amplifiers is a function of numerous factors, including the modulation scheme; a design that works well with GMSK may not be optimum for DQPSK. Efficiencies in the range of 45% to 50% are normal today, and the industry is currently striving to improve this figure beyond 70%. Trade-offs will abound as the 70% efficiency goal is reached. The balance between cost and the ideal efficiency will be difficult to reach without the use of advanced technology.
11.3 MS and BTS—new roads to the ultimate radio Although they perform similar tasks, there are some differences between the radios in base stations and mobiles, and the differences are likely to increase in the future. As we noted in the previous two sections, there is little incentive these days to consider integrated RF functions in GSM radios. This does not mean that new ideas in RF implementations are doomed to be ignored by the industry, for the growing interest in digitized IFs shows that designers are always looking for something better. The feasibility of monolithic GaAs circuits, for example, offers the possibility of
Introduction to GSM technology and implementation
411
increased sensitivity with better control over SNR compromises and the balancing act that is typical in receiver design [13]. Since today’s analog processes are adequate, new digital techniques must substantially enhance performance and reduce costs. Mobile stations are designed to tough cost goals, and network operators shop for smaller base stations with decreased costs per channel. The number of base stations will always be a small fraction of the mobile population. Base stations are designed to pick up the slack the mobiles leave in the air interface: (1) mobiles have tiny, low-power transmitters that require the base station to have very sensitive receivers, and (2) base station transmitters generate much more power than mobiles to compensate for the suboptimal receiving conditions in which users place their handsets. Because base stations do not need to be carried in pockets, all kinds of measures can be employed to make their receivers more sensitive than those in mobiles. Because base sites are expensive and their proliferation over the landscape is less acceptable to the public than in the past, innovations that strive to make GSM systems support smaller and smaller mobiles with fewer base sites are always welcomed by the industry. Increased base station transmitter power can decrease the density of base sites in a region only if there is a corresponding way to increase the base station’s receiver sensitivity. The more sensitive a base station’s receiver becomes, the more likely it is to suffer the effects of interference. Inserting very high Q (highly selective) filters in the receiving path can mitigate this situation, but filters add loss to the path. Cryogenically cooled filters (filters cooled with liquid helium) of very high Q have recently appeared in the market, thus offering filters of extraordinary selectivity without the high loss. Overall system performance can be further improved on the receiver side with smart antennas, which are adaptive arrays that can notch-out interfering signals, or improve the base site’s receiver directivity toward distant mobiles. Another salient example of new innovations in radio technology is the so-called software radio, which is a concept that describes a generalized wideband radio that can be configured, with software, to work in a large number of systems employing diverse modulation and access techniques. The digital IF mentioned in the previous section is an example of this kind of approach. The techniques demand a great deal of performance from baseband functions, primarily in ADCs, which need to drive AGC circuits over a wide range. The increased costs and battery drain that accompany
412
GSM and Personal Communications Handbook
this kind of highly adaptive technique will force the new technologies into the base stations rather than the mobile terminals. There are some advantages in software receivers: (1) any desired bandwidth can be selected, (2) the receiver’s center frequency and phase noise can be controlled, and (3) the I and Q data can be recovered in exact quadrature. Given the increased public intolerance for the proliferation of base sites and antennas, interest will increase in RF multiplexing schemes that allow different base stations to share a single antenna site. A logical extension of this reaction to environmental concerns is the concept of a RF utility, which refers to a base site that can couple many different kinds of RF services (cellular, paging, and PMRs) onto the air interface from different providers of mobile radio services for a fee. This concept takes advantage of one of the most valuable assets a RF service provider may own in a crowded RF landscape: a good radio site.
11.4 Baseband signal processing In this and the following seven sections of this chapter we take a closer look at how a GSM baseband implementation can be realized. The term baseband refers to all signal processing functions related to audio (voice) signals, user rate data, and the radio bits that are transmitted to and received from the radio/IF section. It also refers to data converters and signaling control functions. What are the issues? What is behind such functions as speech codecs, equalizers, Viterbi decoders, converters, and security features? What are typical performance figures and what resources do they require? We reserve space in Sections 11.5 and 11.6 for a closer look at two uniquely digital functions in GSM phones: speech coding and equalization. Although our focus is on a GSM terminal, some of the subjects also apply to the remainder of a GSM network: the base station (signal processing) and the MSC (speech coding and the handling of protocol functions). Among the key technologies that enable and secure the success of wireless services and products, we can easily identify two important ones: semiconductor components and digital signal processing. Together they allow products to shrink in size, cost, and power consumption and still take on new features with no compromises in performance. Silicon technology and other relevant hardware technologies (filters and amplifiers)
Introduction to GSM technology and implementation
413
continue to go through cycles of innovation that allow even lower cost and smaller products. New and powerful algorithms and a growing skill at implementing them efficiently on target hardware, such as digital signal processors (DSPs), provide additional design freedom and product enhancements. The wireless communications industry will continue to drive new technology and clever innovations of older technologies. Figure 11.6 shows the functional partitioning and terms associated with a typical baseband implementation in a GSM phone. A DSP performs layer 1 functions, speech coding, modem, and channel coding and decoding tasks. The so-called mixed signal functions are A/D and D/A conversions. They bring the analog and digital domains together for voice and baseband signals as well as for some control functions. A microprocessor handles GSM protocol functions, general control functions, and the user interface, which is called the man/machine interface (MMI), and which includes the keypad, display, and the user menus. TDMA timing needs to be maintained and supported by a dedicated timer unit. Different kinds of memory hold the microprocessor program and its data (flash ROM), nonvolatile programmable data (EEPROM for equipment identity, phone book, and other more permanent data), active program and data RAM, Battery
Baseband Chip or Chipset
Baseband Line Codec A/D, D/A
Control Processor:
DSP:Baseband Signal Processing, Speech Coding
Protocol Stack
PLL RF Control D/A, A/D
User Interface
Peripherals Control
Memory: RAM Progr. ROM Data ROM
Display
Keypad
Cache Off-chip
Analog Baseband
Addtl. LOGIC TDMA Timer
On-chip
Voice Codec A/D, D/A G. 711
SIM Card
Radio Transceiver Unit
Figure 11.6 Functional block diagram of the digital part of a GSM mobile station.
414
GSM and Personal Communications Handbook
and, finally, the DSP program and its own data, which are typically located on the same chip with the DSP itself for fast access. In the block diagram shown in Figure 11.1, we saw that many functions were needed to provide the speech and signal processing in both the transmitting and receiving directions. Flexibility, gate count, and power consumption let today’s implementation of most of these functions appear in software running on DSPs. All the functions share a single processor, which has to be dimensioned according to constraints on processing power and the available memory. The software modules that demand most of the processor power in GSM are (1) the modem (equalizer), (2) encryption, (3) speech encoding and synthesis, and (4) channel encoding and decoding. The data handling procedures for GSM data and fax services are performed in the place of voice coding functions, and each has its own channel coding mechanism. The processor load for some of the functions can be reduced with the use of hardware accelerators that carry out nonstandard, operation-intensive functions or functions that have to be repeated many times for certain algorithms. Examples are accelerators used for the Viterbi algorithm (equalizer and decoder) and for the ciphering algorithm (encryption engine). Input data are handed over to an accelerator. When the computations are completed, the results (engine output) are available in memories or registers. All the signal processing functions in a full-rate GSM phone (modem, channel coding and decoding, speech processing, etc.) require a certain number of instructions per second. In the processor world the unit of measure for these is expressed in units of millions of instructions per second (mips), and the mips figure for the functions listed earlier are different for each type of DSP, the DSP’s instruction set, and the use of accelerators. The range for a full-rate GSM implementation—using typical hardware accelerators—is around 13 to 20 mips. If we assume one instruction per clock cycle, then this means that a DSP clock speed of 26 MHz is required. This is twice the 13-MHz reference clock frequency from which all GSM timing is derived (see Section 11.10). Furthermore, the software in object code will require a certain amount of program and data memory space as well as someRAM. Direct memory access with no-wait states is secured by reserving a certain amount of program and data memory “on-chip” with the DSP. This ensures that the crucial real-time operations are carried out very quickly.
Introduction to GSM technology and implementation
415
11.5 Speech coding and speech quality in GSM The purpose of speech coding in mobile radio is to reduce the data rate required to transmit speech information from the transmitter to the receiver with minimum adverse impact on the perceived voice quality. Lower data rates can be achieved at the expense of voice quality. Efficient speech coding without sacrificing quality, similar to radio design, is an art of its own. It holds an important place in the world of wireless digital communications. An abundance of speech codecs is available for “wireless” use, each of which features different methods, algorithms, complexities, and performance. The interested reader is referred to [14–17] for additional information on the subject. 11.5.1
Speech coding tutorial
There are three ways to encode speech: waveform coding, parametric coding, and hybrid coding [18,19]. Parametric coders are strikingly different from waveform coders in both performance and construction. Whereas waveform coders can be viewed as elaborate variations on simple A/D converters, parametric coders analyze the input speech signal within strict bounds of, and with full regard for, the processes that created the sounds in the first place. Parametric coders are extremely efficient at removing redundancy from human speech, and offer good voice quality at very low data rates. As a generic introduction into the subject, let us think of ways to transmit acoustic information over an imaginary transmission line, say, a pair of simple telephones connected by a twisted pair of copper wires. One simple way is a very direct one; let us look at an example. The discovery of some new and interesting chords may motivate a guitar player to call up his friend and play the tune to her over the a landline phone, or—why not?—over his brand new PCS phone. In this way the musician’s friend on the other side of the connection is able to listen to the chords instantaneously, but with some restriction in sound quality due to the limited bandwidth in the phone system. Now, let’s consider a more indirect way. Our guitar player might ask his friend to use her own guitar as he instructs her exactly where to place her fingers and otherwise manipulate the strings to duplicate the sounds that he discovered for himself earlier in the
416
GSM and Personal Communications Handbook
day. The result of this more indirect approach is that the original sounds are reproduced, after some delay, with better fidelity than if the sounds were transmitted directly over the phone lines. A similar indirect approach is followed in digital phones. The instructions for recreating the sounds in the near end are transmitted to the far end where some apparatus is placed that can transform the instructions back into the original sounds. Confining our attention to human speech, low-frequency sound (3.4 kHz), such as the low-pass filtered human voice, can be linearly digitized (time and amplitude quantization) at 8 Ksps with 8-bit resolution (A-law or µ-law). This technique is often referred to as pulse code modulation (PCM), which yields a 64-Kbps bit stream, which is the rate used on time-multiplexed digital wireline transmissions such as ISDN. It achieves the expected, and quite adequate, high-quality speech that serves as a reference today. For wireless applications in which the bandwidth for such a high data rate is not available, compromises must be made as additional signal analysis is applied to reduce the redundancy inherent in PCM data. The three main categories of speech processing—(1) waveform coders, (2) vocoders or parametric codecs, and (3) hybrid codecs—are discussed below [15]. 11.5.1.1
Waveform coders
Waveform coders [18,19] work in the time or frequency domain and exploit properties of the source signal waveform (spectral envelope/harmonics/pitch) through methods such as short-term correlation, for example, by the linear prediction (LP) method, which attempt to predict waveform samples from the values of previous samples. There are many kinds of waveform coders [18,20]. Some derive their codes by examining the input speech in the frequency domain, and others do so in the time domain. Waveform coders tend to be relatively simple, take minimum advantage of the redundancy in human speech, and are not particularly efficient at reducing the data rate on the radio channel. Though their data rates tend to be high, and they are not very efficient users of radio channels, waveform coders exhibit excellent voice quality and are tolerant of background noise. Many can even encode music and other sounds that are not of human origin. They work independently from how the signal was generated and attempt a close reconstruction of the original signal. Different varieties work in various ways in both the time and frequency domains. In the frequency domain, for example, sub-band coding
Introduction to GSM technology and implementation
417
schemes split the speech signal into sub-bands that are coded independently. The coding expenditure in terms of bits per second is related to, and in some way proportional to, the energy in the particular band. Waveform codecs have low complexity and, in general, use relatively high bit rates (above 16 Kbps). Linear PCM (64 Kbps) represents the simplest form of waveform coding. ADPCM (at 32 Kbps, CCITT/ITU G.721 standard), which is used in some wireless access systems (DECT and PHS), is a more complex algorithm that still carries a relatively high rate waveform. There is also a variable rate ADPCM, which operates at 16, 24, 32, and 40 Kbps (CCITT/ITU G.726/727). 11.5.1.2
Vocoders
Vocoders (voice coders) are parametric coders that model the voice signal generation in the human vocal tract through a time-varying filter, which gets “excited” in order to produce sounds. This modulating filter is characterized by a mathematical polynomial description through poles and coefficients. The approach is to model the human vocal tract (acoustic filter), which gets excited through a source of energy (vocal cords). This source can be described through two main characteristics of speech: loudness (amplitude), and pitch (frequency). Human speech consists of two types of sounds: (1) voiced sounds (vowels) with a more regular and periodic structure (low frequency pitch, higher energy, and longer duration) and (2) unvoiced sounds (consonants) with a less predictable noise-like characteristic of lower energy that is spread over the whole spectrum. Voiced sounds come from the vocal cords, which form an oscillator with lots of harmonic distortion. A human speaker can change the rather simple oscillations from the vocal cords from one kind of voiced sound to another (different vowel sounds) by changing the shape of a complicated filter, which is the vocal tract. The vocal tract includes the throat, the soft pallet, the hard pallet, the nasal cavity, the oral cavity (including the tongue), and the mouth itself. A human speaker can form dozens of speech sounds with his vocal tract “filter,” but all human vocal tracts are slightly different from each other; each vocal tract has its own presets, which is why we can recognize the identity of a speaker by merely listening to him on the phone. Some talented people can make themselves sound like other people by carefully manipulating their vocal tract dimensions to mimic the vocal tract filter settings of those whose voices they choose to imitate.
418
GSM and Personal Communications Handbook
Unvoiced sounds are the hissing noises that make the consonant sounds, such as the “t” in “top,” the “s” in “sit,” and the “sh” in “shall.” The noise is created by blowing air over or through a constriction to the air flow, for example, between the teeth or between the tip of the tongue and the hard pallet. These unvoiced sounds can be altered by the same vocal tract filter to which the voiced sounds are subjected, which is why we can whisper whole sentences to each other without using voiced sounds, and we can even recognize who is whispering to us in the dark. In the speech coding process, vocoders analyze the original speech signal regarding its contents (voiced or unvoiced, gain, pitch and filter coefficients). The coder assembles a certain set of information for transmission: ◗ Filter coefficients; ◗ Indication of voiced or unvoiced speech; ◗ Gain/loudness values or parameter; ◗ Pitch information for voiced speech.
Reproduction of such sounds in the receiving speech synthesizer, as well as in the loop filter in the encoder, has to be different for different sets of voiced and unvoiced speech samples. Voiced speech will be reproduced by a filter excitation through periodic or regular (glottal) pulses. The pitch of the sounds can be represented by the distance (in the time domain) between the pulses. For unvoiced sounds, the excitation of the filter is random (white) noise. See Figure 11.7 for a depiction of speech synthesis. These kinds of coders are excellent at moving intelligible voice over narrow channels at astonishingly low bit rates. The disadvantages are that parametric coders are much more complex than even the most elaborate waveform coders, and they can only handle human speech sounds. Vocoders operate at low bit rates (down to 2.4 Kbps) and, though their voice reproductions are fully intelligible, they tend to sound rather synthetic. The problem with this simple vocoder approach is that it resolves only very short intervals between voiced and unvoiced sounds. So, in the gray area between different sounds, the determination of the characteristics of the speech may not be adequate. In addition, interactions between the excitation and the filter (sound source and vocal tract) are disregarded.
Introduction to GSM technology and implementation
419
Speech Information from Encoder Pitch
Voiced/ Unvoiced
Peridodic Pulse Excitation
Gain
Filter Coefficients
Time Variant Synthesis Filter Model
Reconstructed Speech Samples
White Noise Excitation Speech Synthesis at Receiver
Figure 11.7
Speech synthesis with vocoders.
Thus, the use of vocoders is not recommended for normal telephony, but they find wide use in professional radio systems such as in public safety applications. 11.5.1.3
Hybrid coders
Hybrid coders, which are generally known as analysis by synthesis (AbS) codecs, introduce a more distinct analysis of the excitation source. They transmit either a residual excitation model, or an excitation pulse shaping, sizing, and spacing model, or a codebook-oriented selection of excitation pulses (code excitation linear prediction [CELP]). They fill the gap between the two methods described earlier. They make use of linear prediction (waveform coding), and they replace the vocoder method, which simply distinguishes voiced and unvoiced sounds, with a better excitation signal processing scheme. Hybrid coders try to minimize the error between the input speech waveform and the reconstructed speech waveform by finding the ideal excitation signal. Figure 11.8 illustrates a simple but imaginary hybrid vocoder. Though the encoding and decoding processes in speech coders are usually accomplished within the same physical device, the illustration in Figure 11.8 has
420
GSM and Personal Communications Handbook
speech level LPF
Balanced Modulator
Filter
voiced / unvoiced
Speech Analyzer pitch counter Pulse Generator
Noise Generator
Filter Control
Difference
Encoder channel Decoder
speech
Filter
Balanced Modulator
level
voiced / unvoiced
Pulse Generator
Figure 11.8
Noise Generator
pitch
Simple hybrid speech coding example.
the two processes separated for clarity. The encoding function is in the top half of the figure, and the decoding function is in the bottom half of the figure. The two halves are separated by the radio channel as indicated in the oval. The imaginary speech coder regards human voice, in short frames of typically 20 ms, as either voiced vowel sounds, or unvoiced hissing sounds. The speech analyzer in Figure 11.8 separates voiced from unvoiced sounds, determines the level of the voice, estimates the fundamental pitch frequency in voiced sounds and sends data representing these three simple voice characteristics over the channel. The voiced or unvoiced
Introduction to GSM technology and implementation
421
parameter can be represented by a single bit. A filter in the encoder gets some adjustment instructions from a block that compares the filter’s output with the actual speech input, thus closing a control loop that maintains the filter’s output as a faithful copy of the original speech. If the control loop’s information is also sent over the channel to an identical filter (in the receiver), then the filter settings will alter the properly adjusted voiced or unvoiced sources such that a copy of the original speech leaves the filter. Consider, again, our guitar player teaching his friend some new chords over the telephone. He can either play the chords into the telephone’s mouthpiece and hope the connection is clear enough that the music can be discerned in the student’s ear (waveform coding), or he can tell his friend which chords to play and how to play them, (parametric coding). The reader may note that all the functional blocks in the lower receive portion of Figure 11.8 also exit in the upper portion of the drawing. This serves to illustrate how functional blocks can be used for two purposes, which is not particularly difficult if all the functions are actually data manipulations in a DSP. As we explained before in our example of a hybrid coder, the error minimization is already done in the coder, which passes the excitation recursively through the synthesis filter that already has some presets (AbS). The speech signal gets chopped up into short frames of 20 ms, over which the analysis is carried out. Once the ideal excitation and filter parameters are found, they are transmitted to the synthesizer. In CELP codecs, a codeword within a codebook is indexed by a so-called vector (a pointer). Such vectors (the only information that is actually transmitted) point to excitation values stored in the codebook in the receiver (decoder). This procedure reduces the number of bits needed to transmit the excitation information. However, codebook searches in CELP codecs are very complex tasks that require relatively high processing power. Codebooks also use up quite some nonvolatile memory (ROM tables), which needs to reside with the signal processor. The whole transmitted speech information frame consists of synthesis filter parameters (eventually including long-term prediction and pitch information) and the appropriate codebook excitation vectors for the speech synthesizer at the other end. At the speech synthesizer, an excitation of the synthesizer is typically performed with nonzero pulses at a rate of 4 to 10 pulses every 5 ms. This implies that for a 20-ms speech frame, up to 40 excitation pulses (positions and amplitudes) must be found by the analyzer and
422
GSM and Personal Communications Handbook
generated in the synthesizer. Hybrid codecs, which also appear in many variations and combinations, typically work at net bit rates that are somewhere between the ones known for vocoders and waveform codecs, which puts them between 4 and 16 Kbps. Some speech coders, such as the Qualcomm Q5414 variable rate vocoder (which is a hybrid codec [14]), can actually reduce their output data rates in accordance with the lack of need for speech coding work. 11.5.2
Speech quality
The line quality of a digital communication link is, to a substantial degree, determined by the speech codec. Other effects including delays, echoes, interference, and interruptions, which do not fall within the speech codec’s responsibility, may have an impact on the perceived voice quality. The speech coder’s stability, or its susceptibility, remains a determining factor that will be tempered by the channel coding regime and modem performance. Disturbing effects may come from the network (echo cancellation, round-trip delays), environmental influences (background noise), and signal reception conditions (interference and fading). The perceived speech quality is also affected by the terminal’s acoustic design. Voice activity detection (VAD), which is used for speech-dependent transmission (DTX), may introduce annoying clipping effects. One network-related factor influencing end-to-end speech quality is tandeming. Conversion of GSM speech to PCM, or even further to analog signals, and back to GSM speech hurts speech quality, because distortions are introduced that increase with each conversion. Tandem-free operation avoids transcoding as much as possible. In mobile-to-mobile calls, the encoded speech can be transferred directly at 13 Kbps without any conversions to PCM within the PLMN. Another annoying effect on perceived speech quality is the delays and echoes that cannot be compensated for in dedicated echo canceling units in a GSM MSC. On the matter of quality, how do we measure a speech codec’s performance? Acoustic measurements on the terminal side are performed with an artificial mouth and ear [21]. The relevant parameters are (1) loss, (2) sending and receive loudness ratios (SLR and RLR), (3) frequency response, (4) acoustic SNR, (5) sidetone, (6) out-of-band signals, and (7) crosstalk. For a GSM network, TS 03.50 specifies the transmission performance requirements [22].
Introduction to GSM technology and implementation
423
Then, categorizing speech codecs is, in general, achieved with tests performed by trained listeners. So-called mean opinion score (MOS) values are obtained for different speech coding algorithms by subjective auditory tests performed by trained human listeners under controlled conditions. The conditions, or error profiles (EPs), include the static carrier-tointerference ratio (C/I) at the radio receiver, for example, at 13, 10, 7, and 4 dB, or 0%, 5%, 8%, and 13% for average gross bit error rates and dynamic radio path conditions (fading, Doppler, and frequency hopping). MOS figures can rank between 0 and 5. An MOS of 4.0 already indicates relatively high perceived wireline quality (as achieved through PCM), which is often referred to as toll quality (the transmission quality maintained between toll centers). There is a tight relationship between speech codec rate, achievable quality, and the complexity of speech codecs. The better the perceived quality and/or the lower the net bit rate, the more complex (mips and memory requirements) a speech codec implementation will be. Other important quality aspects relate to robustness (resistance to bit errors and burst errors) and to the efforts that must be included in channel coding and decoding. The better the protection through channel coding, the better the speech quality. 11.5.3 tones
DTMF and signaling
Dual-tone multifrequency (DTMF) is commonly used in telecommunications systems to transmit user data or application-specific information (digits 0–9 and the characters A, B, C, D, *, and #) with audible tones generated from two audio sources selected from a four-by-four matrix. Applications include the submission of a PIN, a credit card number, or other access and control codes. DTMF is used for accessing and controlling answering machines and voice mail services, as well as automatic teller services. In GSM, DTMF is handled in one of two ways that depend on the origin of the tones. The GSM speech codec does not handle DTMF tones (monotonous sine waves) too well. It introduces a non-negligible amount of distortion. So, in the uplink direction (MS to network), the DTMF digits are delivered through signaling. A DTMF message is sent over a signaling channel (FACCH) for each digit. This is a normal application case. In the downlink direction (network to MS), the tones can be
424
GSM and Personal Communications Handbook
actually carried in the speech channel, thus treating them exactly like voice signals, even though the speech codec would introduce distortion. However, so far the application of a GSM MS having to decode DTMF tones has not appeared and also was not anticipated. Still, a GSM speech codec has to be transparent, with a minimum of audible distortion added to DTMF so that an analog DTMF decoder still could detect the tones. The same requirement for such transparency is true for other acoustic signaling tones from a network. Some of these other signals are the familiar ringing sound (ring-back indicator), the busy line indicator, and the warning signal. 11.5.4 GSM full-rate speech coding
The GSM full-rate speech coding function is described in the 06 series of the GSM specifications for the digital cellular telecommunications system [23]. The specifications are the only reference that can be given for engineering purposes. More complete treatments of the subject can be found in [2,24]. GSM full-rate speech coding makes use of a RPE/LTP combination of processes. This combines linear predictive coding (LPC), which exploits short-term correlation in the speech signal, with long-term correlation (through LTP). The residual signal, which excites the vocal tract model in the receiver, consists of a set of regular pulses. The RPE-LTP contender received the highest average MOS score among the six final proponents for the GSM full-rate speech codec. All the codecs were tested with seven languages under various transmission conditions (three different input levels and bit error rates, one and two transcodings, and two environmental noise forms) [25]. It is interesting to note that the actual language spoken has an effect on the speech codec efficiency. This phenomenon is true for many lowcomplexity predictive vocoders, and it has to be taken into account when qualifying improved codecs for future use in systems deployed in many countries. There is also a distinction between female and male voices, which some codecs handle one or the other better. 11.5.5 GSM half-rate speech coding
The GSM half-rate speech processing function is described in the 06.20 series of the GSM specifications (ETS 300 581 series) [4]. It makes use of a
Introduction to GSM technology and implementation
425
5.6-Kbps CELP/VSELP combination of speech codecs. The net 5.6-Kbps data are channel coded (GSM 05.03) with a different coding scheme than that used for GSM full-rate channels. Because each bit is more important in a low-rate process, a higher complexity convolutional coder with more memory is used. That is, instead of a constraint length of five (k = 5), the half-rate data are protected through a constraint length of seven (k = 7). The convolutional coding, which is applied to the most important class 1 bits, is combined with a puncturing mechanism. If we include the cyclic code parity bits and the tail bits, then a gross rate of 228 bits per 20-ms frame (11.4 Kbps) is used for block diagonal interleaving over four bursts. This represents exactly half the gross rate of that used for full-rate traffic channels (22.8 Kbps). The motivation for this speech codec is that it allows for the use of one time slot in every other TDMA frame in one dedicated traffic channel, which doubles the physical channel capacity. This is regarded as an important feature by GSM network operators. Because cellular spectrum is a rare resource, and some densely populated areas already have capacity problems during peak hours, half-rate would relieve this situation. However, as nothing comes to us for free, there are some drawbacks. First, only a significant number (30% to 50%) of half-rate users in a network can generate noticeable relief in capacity inside bottleneck areas. This is difficult to achieve with an existing user base because the exchange/replacement period for terminals is between 18 and 36 months. Second, half-rate terminals need to be able to operate in full-rate environments. This is because a network may not be completely covered by half-rate capability, or the user may want to roam within other fullrate networks. Such dual-voice-codec terminals would be slightly more complex and slightly more expensive. Third and finally, the speech quality offered by half-rate vocoders is regarded as lower than full-rate in MOS trials. The MOS figures fall in the range of 3.0 and a comfortable operating region in terms of error profiles was more limited. The presence of background noise and/or tandeming conditions can make half-rate coded voice almost unintelligible. 11.5.6 GSM enhanced full-rate speech coding
The GSM EFR speech processing function is described in the 06.50/06.60/ 06.80 series of the GSM specifications (ETS 300 723–300 730) [5]. The
426
GSM and Personal Communications Handbook
coding scheme was adopted by the American ANSI T1P1 committee as a higher quality speech codec for PCS 1900 systems based on the GSM platform. It was also adopted within ETSI through GSM Phase 2, thus becoming a European Telecommunications Standard (ETS). The GSM EFR is a so-called algebraic code(-book) excited linear prediction coder (ACELP). This indicates the use of an algebraic codebook, in which an algebraic code is used to arrange or populate the excitation vectors within the codebook. This makes the search algorithm more efficient. The EFR generates speech compression information at the same bit rate as the full-rate process does (13 Kbps), and the coding and interleaving schemes are identical. The EFR was finally accepted after some debate on whether channel coding should be adapted to better fit the EFR codec and to further improve performance in low C/I conditions. This alternative was eventually rejected because some manufacturers had already committed to hardware implementations. The perceived voice quality achieved through the EFR is much better than the full-rate scheme. It has been tested thoroughly, and the MOS figures are above 4.0 in office and street noise environments. It is said by some to even outperform the ADPCM in the presence of background noise. The use of EFR codecs in established GSM markets, where there is a large population of phones already deployed, only makes sense in combination with the GSM full-rate standard as a fallback in both the networks and the terminals. In newly established networks, such as in PCS 1900 networks in North America, the EFR is seen as the main standard. It turns out that all the world’s digital cellular systems have recently specified and deployed improved codecs of similar quality to GSM’s EFR. For example, the ACELP has replaced the VSELP in some of the IS-136 systems, and the 13.3-Kbps coder and enhanced variable rate codec (EVRC) have replaced the 8-Kbps process in the IS-95 systems. Time brings steady improvements to speech coding technology; better quality speech can be realized with the same data rates of only a few years ago. Further improvements are sure to be proposed in the future: the advanced multirate (AMR) proposal may eventually compete with some wideband (50-Hz to 7-kHz) voice coders for attention in the GSM community. The former is proposed as a way to get half-rate service deployed, and the latter are proposed to allow GSM to compete directly with wireline services. Because GSM systems exist next to and compete with IS-136 and IS-95 systems in North America, the Americans have had to push the EFR specification
Introduction to GSM technology and implementation
427
through their own T1P1 standards body rather than wait for the slower ETSI process. 11.5.7 Complexity comparison FR-HR-EFR
We have to consider the required resources before implementing a speech codec in a GSM mobile phone or in a network component. A certain amount of signal processing power must be reserved to make sure that all the real-time coding and decoding functions are carried out in time. Program and data memory resources must be provided with adequate access time, and temporarily accessible memory (RAM) has to be available. Table 11.1 compares optimized full-rate, half-rate, and enhanced full-rate speech codec implementations. These are generic figures that take into account assembler programmed, standard 16-bit, fixed-point DSPs that are used in wireless applications. Different DSPs perform differently for some operations and in some implementations. There is also a human factor in that all the figures are also programmer dependent. For half-rate codecs, note that due to the higher constraint length in the convolutional coder (k = 7), the peak decoder-complexity increases. This is, however, partially compensated by the longer receiving intervals (double) through the half-rate operation. 11.5.8 The future for GSM speech coding
Most GSM network operators find themselves in very competitive environments with other operators that use digital and analog, and fixed-line
Table 11.1
Comparison of Voice Coding Complexities Speech Codec
Maximum mips
Program and Data ROM
RAM
GSM FR
2.5–4.5
4–6 kWords
1–2 kWords
GSM HR
17.5–22
16–20 kWords
~ 5 kWords
GSM EFR
17–22
15–20 kWords
~ 5 kWords
428
GSM and Personal Communications Handbook
or wireless, technology. Because voice telephony remains the most widely used service, audio and speech transmission quality can be made a major differentiator in the market for both networks and terminals. With traditional full-rate technology, GSM offers more or less adequate quality at a relatively high transmission rate at the radio interface. But other options are being explored. As speech processing technology matures and improves, lower bit rates may provide even better quality. This reduction in net bit rates can be used to increase robustness through channel coding, or to increase capacity through high-quality half-rate vocoders. The gap between full rate and half rate may eventually fade away: variable rate vocoders can fill the gap and allow on-demand scaling of speech data rates and transmission rates. When available, when necessary (bad radio conditions), or when subscribing to a premium service, the user gets a high-quality, full-rate channel with a robust wireline-like transmission quality. When radio conditions are good and/or more capacity is needed, the rate falls back to a lower bandwidth that still provides adequate quality. ETSI has responded to industry requirements for open standards on the next-generation speech codecs. The specifications within SMG 11 for an AMR codec are being elaborated. The GSM specifications describe the entire PLMN and interworking protocols, so there is much more than just a particular coding algorithm to be worked out and agreed on. There are also a number of signaling and control issues, channel coding, and testing requirements to be considered. The introduction of the AMR codec was requested by the network operator community to be ready shortly after the turn of the millennium, and the standards bodies are obliged to respond. Tandem-free operation (TFO) refers to mechanisms that avoid multiple speech coding and decoding steps such as when, for example, a mobile phone asks for a voice channel with another mobile phone within the same PLMN. The lack of TFO in this case would mean that the encoded speech is converted into the wireline system’s PCM scheme at one BSC, and the PCM code is converted back into the wireless scheme through a codec in another BSC; each conversion diminishes the speech quality. TFO is being tackled by a working group within ETSI SMG 11. Definitions and specifications have to be elaborated in order to provide for the appropriate control mechanisms. This is necessary in order to engage TFO whenever possible in calls within a GSM PLMN as well as among GSM PLMNs.
Introduction to GSM technology and implementation 11.5.9
429
Speech coding and…
There are some other matters that need to be considered when processing speech and audio/acoustics. Two important ones are hands-free operation and voice recognition. 11.5.9.1
Hands-free operation
Though the practice by automobile drivers is increasingly seen as lacking in social responsibility, mobile telephones are used while on the move. The use of a hand-portable phone in a car is supported with a connection to a hands-free kit. This kit extends four interfaces within the terminal: the microphone, the speaker, the battery, and the antenna. The microphone used in a car kit is often installed close to the speaker’s mouth, perhaps just beneath the inside rearview mirror. It is small and has a directional reception characteristic. A loudspeaker is placed wherever there is room to spare. A connection to the car’s battery saves the terminal’s own battery capacity, and even allows for the possibility of charging the phone’s battery when the kit is used. The outside antenna included with the kits provides greatly improved radio signal reception and transmission. As far as the phone is concerned, a car is a Faraday cage that isolates a radio transceiver from the host cell sites. With only the internal (built-in) antenna of a hand-portable unit, the reception is impaired because the power control mechanism drives the radio’s transmitting power level to the highest power levels. Thus, the external antenna affords improved signal reception, longer battery life, and better voice quality, and reduces radio interference with the car’s electronic systems and other potential influences inside the car. There is growing evidence that a driver using a hands-free kit to engage himself in a phone conversation is sufficiently distracted from the driving task to pose a safety concern. Some countries do not allow drivers to talk on the phone without the hands-free feature, and the prudent automobile operator may want to consider restricting the use of the hands-free feature to passengers. Hands-free operation takes place within a complex acoustic system consisting of a loudspeaker, a human speaker, a “room” or chamber, and a microphone. Figure 11.9 shows the acoustic system at the near end. Without any further treatment, the incoming audio signal from the far end would emerge from the loudspeaker and be received by the
430
GSM and Personal Communications Handbook
Room / Car Loudspeaker
Receive Path from far end
NOISE
Transmit Path to far end Microphone
Speaker
near end Figure 11.9
Hands-free environment.
microphone through a direct path as well as through some additional indirect reflections that depend on the “room.” Because there is not much attenuation in the acoustic link, an echo would result (perhaps with some feedback) at far end. Under the best circumstances, the far end speaker would hear herself with an echo. Under the worst circumstances, the feedback path would quickly build up an annoying howling sound. Some tricks are employed to prevent these annoying effects. Two of the countermeasures are echo suppression and echo cancellation. E c h o s u p p r e s s i o n Echo suppression means that the inactive audio path gets attenuated while the active path is left open as in normal nonhands-free operation. Echo suppression introduces a half-duplex flavor to the conversation mode; only one path is active at a time. Echo suppression requires some kind of VAD at the near end for the signals from both ends. Voice activity detection can be quite complex, especially in the presence of background noise so common in moving vehicles. Many VAD schemes exhibit some kind of annoying clipping effect that is generated when the voice is clipped off when a certain energy threshold is achieved. Because the speech coders in GSM terminals need to support VAD and DTX anyway, the information is present for use in hands-free logic: (1) conflicts arise when both ends speak at the same time, (2) an idle status is
Introduction to GSM technology and implementation
431
resumed when there is no speech from either end, and (3) hangover periods are introduced to cover the time when a speaker resumes talking after a short pause. All this is handled in the hands-free control logic, which performs its tasks through relatively simple digital signal processing. Attenuation and volume control can also be carried out digitally just after, or immediately before, the PCM conversion in the handset. These kinds of tasks consume something in the range of 1 mips or less in a standard DSP. The memory requirements for program, data storage, and RAM space are relatively modest so that echo suppression is an attractive solution for low-cost implementations. It is not a very elegant solution. At the time of this writing, echo suppression was the only technique widely employed by mobile terminal and accessory manufacturers. The hands-free control logic can add lots of perceived value to otherwise similar products; some are much better than others. In general, echo suppression provides adequate quality, especially when some speaking discipline is applied at both ends. Figure 11.10 shows the concept of echo suppression for hands-free operation. E c h o c a n c e l l a t i o n A more elegant, but much more complicated and costly solution is echo cancellation. Rather than abruptly clipping off voice signals and ruthlessly attenuating audio paths, echo cancellation maintains a full-duplex link between both ends. How can this be achieved without the annoying echo effects? Complex dynamic or adaptive echo Volume Control
Receive Path from far end
Room / Car Loudspeaker NOISE
VAD
VAD
Transmit Path to far end
HandsFree Control Logic
Volume Control
Microphone
Speaker
near end
Figure 11.10
Echo suppression for hands-free operation.
432
GSM and Personal Communications Handbook
cancellers work with a system model that describes the closed room system. The concept is similar to the more static echo cancellers used in fixed networks and switches. In particular, all of the acoustic signal paths between the loudspeaker and the microphone form an acoustic system. Such a system can be modeled through a complex time-variant filter function, which is used to calculate and then subtract out the echo signal that would have occurred if the acoustic system were left alone. This echo signal resembles the original far end signal that was modeled or shaped by the acoustic system on its way to the far end. The concept of adaptive echo cancellation is shown in Figure 11.11. For simplification we will use the terms of the z-transformation for time-discrete signals. The z-transformation does not apply to the analog time-continuous audio signals of the “room” system. In Figure 11.9, the incoming signal, X, is shaped through the “room” filter, H(p). The system response estimator calculates and maintains this digital filter function, S’(z), which models H(p). The calculated echo signal, X’(z), is subtracted from the return signal, Y(z). The problem with the implementation in adaptive echo cancellers for hands-free operation is that the ideal processing solution has not been found yet. Trade-offs have to be made in complexity, performance, and adaptivity (rate). Especially when the echo mechanism is not a static one, but rather undergoes changes (additional noise, movements, and when Room / Car Receive Path from far end
Loudspeaker
X (z)
NOISE
Adaptive System Response Estimator & Compensation Filter S’ (z)
X’ (z) Y’ (z) Transmit Path to far end
−
+
+
H (p)
Y (z)
Microphone
Speaker
near end
Figure 11.11
Concept of echo cancellation.
Introduction to GSM technology and implementation
433
both parties are speaking from both ends), the ideal path can get lost very quickly. Digital filters tend to have a dependency on their complexity and their training period, which can cause trouble in some cases. Even though there are already some good solutions, which require us to add another DSP into the system, research in this area is ongoing and will certainly yield better solutions in the future. The solutions need to have adequate performance with reasonable cost and processing effort. Technical implementations require something in the range of 20 mips and above, and are realized in a DSP located within the handset cradle (which is connected to the hands-free speaker and microphone) or in the handset itself. 11.5.9.2
Voic e recognition
Voice recognition technology, similar to echo cancellation, has received a lot of attention in the communications industry. For the personal communications world, this means that terminal functions become accessible through simply “telling” the device what to do. Simply say, “Call Mom,” and the device dials your home number so that you can check what’s for dinner. Voice recognition works either with a trained set of personalized commands, which is a less complex implementation that requires some upfront tuning and storage work, for example, for the user’s 16 favorite wishes, or with a generic and much more complex approach which reads all characters and words. A similar approach is evident in character and handwriting recognition on touch-sensitive screens where a computer recognizes handwriting which it converts into machine text and characters.
11.6
Equalizers
Under ideal situations a transmitted symbol should arrive at the receiver, greatly attenuated, but undistorted and occupying only its intended time interval. Unhappily, such is seldom the case in mobile radio. When one symbol is so distorted that it occupies the time reserved for other symbols, we have intersymbol interference (ISI). The situation is made even worse in GSM as the GSM transmitter itself contributes is own ISI. There are two contributions to ISI in the GSM system: (1) uncontrolled ISI from the radio channel and (2) controlled and deliberate ISI from the transmitter’s partial response modulator.
434
GSM and Personal Communications Handbook
The reader is cautioned that a complete treatment of the causes of, and the solutions to, ISI are not particularly intuitive. First, the mathematics involves vector quantities rather than scalar ones. Add the scalar quantities, 5W and 10W, to get an intuitive 15W. One dollar plus another dollar is two dollars. Vector numbers (also called complex numbers) have both an amount and a direction. One kilometer to the east (90o) plus another kilometer to the east equals 2 km to the east; but 1 km to the east (90o) plus 1 km to the north (0o) equals 1.414 km to the northeast (45o), a not particularly intuitive result without some personal reflection. Engineers work with complex numbers so often that they eventually gain considerable intuitive sense of their manipulations. Further, the mobile radio channel’s impairments, fading, multipath, and Doppler spread, are tempered by the modulation scheme in subtle ways that require sophisticated simulations on computers in order to deduce the BER performance or the probability of an outage. Finally, the effects of ISI are influenced by the effectiveness of signaling techniques. We enlist the relatively simple linear examples in phone lines and our experience with analog FM radio to explain what an equalizer is and bring some appreciation for how this important device deals with the impairments of the radio channel. The engineer who must effectively deal with radio channels and equalizers can turn to the books suggested in Section 11.12 and to his computer. An equalizer can repair some of the damage caused by ISI in a radio channel or in a phone line, and Figure 11.12 shows that the equalizer is the first function in the receiving path after the demodulator. All digital receivers have a decision circuit that constantly passes judgment on the state (e.g., “1” or “0”) of each and every noisy and distorted symbol emerging from the demodulator. The equalizer is inserted between the demodulator and the decision circuit, thus removing some of the ambiguity from the distorted symbol stream. Equalizers are neither free nor do they draw negligible power; they are only used if their cost and burden on the available power can be justified. Most implementations of GSM receivers employ a type of adaptive, nonlinear, maximum likelihood sequence estimation (MLSE), soft output equalizer known as a Viterbi equalizer [26,27]. Many different types of Viterbi equalizers have appeared each differing from the others in details, complexity, and performance [28]. Their details are usually shrouded in secrecy as the different equipment makers seek competitive advantage in the market by improving the performance of their receivers with
Introduction to GSM technology and implementation
435
Input
Demodulator
Equalizer
Decision
De-interleave
Channel Decode
Desired
Actual
Σ Speech Decode
Error
Signaling Decode
Outputs
Figure 11.12
A digital receiver.
equalizer improvements. The secrecy in the market is further compounded by the highly specialized nature of present-day baseband techniques. Because a rigorous and complete treatment of Viterbi equalizers is too specialized for this chapter, we endeavor to ford the river that separates the specialist from the general reader as we describe these devices intuitively with pictures and simple examples. 11.6.1
The problem—ISI
The primary cause of ISI is multipath fading. It is not likely that a radio signal will propagate unimpeded from a transmitter to a receiver through free-space along only one path; the transmitter’s signal typically follows many paths to the receiver. In addition to the direct path, different types of alternative paths are (1) reflected off a structure that is large compared to the wavelength of the signal, (2) scattered off an object that is small compared to the wavelength of the signal, and (3) refracted around the corners of objects. Another cause of ISI is Doppler spread, which is due to the relative motion between a receiver and a transmitter; each path the signal’s
436
GSM and Personal Communications Handbook
wave takes from the transmitter to the receiver experiences an apparent shift in frequency. The signal that finally arrives at the receiver is the sum of many signals with random amplitude (the amount) and phase (the direction). The amplitude of each individual signal changes with the loss on its path and the phase of each individual signal changes with the length of its path. In some instances the sum of all the signals will add constructively, improving the received signal, and at other instances the signals will add destructively degrading the received signal. When a large number of such signals from a transmitter is present at the receiver, each a function of two random variables, amplitude and phase, we have a Rayleigh distribution. This means that we know the probability of encountering a fade of some specific amplitude in a multipath environment. Let’s set up an experiment in a city, or any typical place GSM handsets are likely to be used. If we measure the received signal strength, at a test receiver located anywhere in our city, of the emissions from a small transmitter carried around in a circle of, say, 1-km radius, from the test receiver’s antenna, plot the recorded signal strengths on the vertical, logarithmic axis of a piece of graph paper, then we will get a curve that looks something like that shown in Figure 11.13. The curve shows occasional deep dips, or fades, in the received signal strength. Now, let’s add a sorter to Figure 11.13. A sorter is a set of small boxes arranged on the vertical axis. Each box represents a small range of received signal strengths. As
sorter
−75 dBm
−117 dBm
time
Figure 11.13
Rayleigh fading.
Introduction to GSM technology and implementation
437
we move steadily along the horizontal time axis, we will add a count to whatever box represents the observed signal strength at every instant we stop to record them. When we finally arrive at the end of our time axis we will see that the distribution of counts in the boxes is not symmetric. In the case depicted in Figure 11.13, the box that includes the –75-dBm received signal strength will hold the largest count, the counts in the boxes that represent slightly stronger received signal strengths will be almost as high (decreasing quickly as we examine the contents of boxes representing even stronger received signal strengths), and the boxes representing very weak received signal strengths may have accumulated counts. The counts in the boxes that represent signal strengths below –75 dBm decrease more slowly than counts in the boxes above –75 dBm; the distribution of the counts around the –75-dBm box is not symmetric. The normal (Gaussian) distribution has a symmetric distribution of sorter box counts around some most likely value. This would occur, for example, if we observe the weights of cookies emerging from a machine designed to manufacture cookies of 150g in weight. Most will weigh very nearly 150g, but it is just as likely for one to weigh 140g as another is to weigh 160g. Both processes, the possible received signal strength from a transmitter and the weights of cookies, are random ones, but they are described in different ways and have different distributions of possibilities around some dominant value. Yet another distribution is the Rician distribution that applies when the direct path from the transmitter to the receiver dominates all the other paths. If we move our test transmitter in a straight line away from the test receiver instead of moving it in a circle around the receiver, then the “bumps” in Figure 11.13 will fall as the test transmitter moves further and further away from the receiver. Most of us have heard the effects of multipath fading in analog FM mobile radio. As we carry a portable radio in a large circle around a test receiver, we note that the signal strength varies in a manner depicted in Figure 11.13. We can sometimes move the radio around (over a distance of ½ wavelength, λ/2, which in the typical 900-MHz cellular band is about 15 cm) until the resulting input to the receiver is weak enough—destructive interference—to allow noise to dominate in the speaker; we have thus positioned the radio in a deep fade. If we drive around the test receiver fast enough, our receiver will pass through the fades quickly enough to cause pops in the speaker. The pops are sometimes described as
438
GSM and Personal Communications Handbook
“popcorn” or “cooking bacon” sounds, and are caused by abrupt phase rotations in the carrier-plus-noise phasor that yield unwanted phase modulation. The phase modulation manifests itself as clicks and pops from the FM detector. The physics that causes these amusing events in analog FM mobile radio conspire to cause ISI in digital radio by spreading or smearing the symbols that arrive at a receiver. One bit interval in GSM—in which the payload data symbols are transmitted using a kind of phase modulation (GMSK)—is a mere 3.69 µs, and the typical delay spread (the time interval over which signals following different paths from a transmitter arrive at a receiver) in GSM systems is about 5 µs, which is longer than one bit time [2,27,29]. Because there is only one bit in a symbol, we can equate the bit time with the symbol time. The short symbol time means there are plenty of opportunities for ISI, and the relatively short burst duration and reoccurrence time does not allot us much time to repair the symbol damage in the receiver before the next burst arrives. Figure 11.14 depicts a simple case of ISI, where the time interval, t, between points 4 and 5 could be the interval the original transmitted symbol actually occupies. The short original pulse (symbol) arrives at the receiver spread over eight symbol times. There are many causes of ISI. Multipath fading is one. Shoving pulses (representing symbols for the payload data) through a filter too narrow to
1
2
3
a4
a5
4
5
6
t
Figure 11.14
Intersymbol interference (ISI).
7
8
9
Introduction to GSM technology and implementation
439
accommodate them is another cause. Indeed, the limited bandwidth of the mobile radio channel is the chief constraint on symbol rates in digital radio systems. Narrow filters spread the modulation pulses over time as they round off the edges of the pulses and the mobile radio channel has its own time-varying and largely unknown filter-like influences: Doppler spread and multipath fading. They make the radio channel appear to be a filter the characteristics of which change with time. Because the destructive influence of the mobile channel changes with time, the compensation techniques must be adaptive. A static channel, such as we may find in a long cable, exhibits a fixed influence on the pulses. We can use much simpler nonadaptive and linear techniques to clear ISI in such wellbehaved channels. Equalization is only one of the techniques used to clear the effect of ISI. Two others are diversity and channel coding. All three techniques are used in parallel to counter the effects of ISI. Diversity works on the causes of fading channel impairments; it attacks at least some of the causes of the problem by reducing the depth of fades. Diversity techniques are well known. The use of multiple receiver antennas, or diversity reception, is an example of space diversity, and it works by exploiting the highly unlikely possibility that deep fades, which occur at half wavelength intervals, will be experienced in two antennas, separated by some distance, at the same time. There is 1 chance in 6 that a rolled die will fall with two dots face up, but there is only 1 chance in 36 that two dice will land with two dots face up on both cubes. Frequency hopping is an example of frequency diversity [30]. Diversity measures are considered successful if the duration and depth of fades are reduced to a manageable level. Channel coding is an inoculation technique against ISI. The channel coder adds carefully contrived and clever redundancy to the transmitted data which the receiver can use to detect and, within certain limits, repair the underlying information data [31,32]. The receiver can perform its own repairs or it can ask for repeated transmissions. Though they are not perfect in their ability to repair destroyed data, channel coding techniques have become very powerful in recent years [32–35]. Just as is the case with the voice codec, both the receiver and the transmitter must agree on a common channel coding scheme for the techniques to work. Equalization works on the symptoms rather than the causes of ISI. In one sense, the more simple forms of equalization are compensation techniques; whatever the modulator and the mobile radio channel do to the
440
GSM and Personal Communications Handbook
original transmitted data to cause ISI, equalization does the opposite to compensate for the influences [36]. Another form of equalization is the MLSE that is used in GSM. This form tries to figure out what the transmitter’s original information (input) bits were from the received (distorted) symbols (output) together with an estimate of the channel that caused the distortions. Equalizers “undo” the observed ISI in the receiver in widely different ways, and the most advanced techniques use signal processing techniques within the baseband portion of the receiver to perform their work. The difference between the three ISI countermeasures, diversity, channel coding, and equalization, are not always clear. The CDMA system mentioned in Chapter 3 uses an elegant combination of all three methods. The three methods come together at one point in the RAKE receiver [37,38]. The RAKE receiver itself can be viewed as an elaborate equalizer that takes advantage of time diversity to make the system work. Time diversity is a form of multipath fading in which the paths differ so much in time that appropriate techniques can be employed to actually improve overall performance. A salient characteristic of the type of GMSK employed in GSM is the relatively low filter bandwidth, B, and bit duration, T, product, which is the BT product: BT = 0.3. The low BT product implies a gradual change of the transmitted carrier from one symbol to the next in such a way as to spread each symbol change over several symbol periods. A bit change is spread over three bit times when BT = 0.3. Were the BT infinite, a modulation bit’s influence would be confined to only one bit time, but the resulting signal would be simple MSK, which responds to modulating symbol changes through abrupt phase shifts in the transmitter. Figure 11.15 illustrates the difference between MSK and GMSK. The data applied to the modulator and the resulting phase change created by the modulator are plotted over time for both the MSK case (the fine line with sharp corners) and the GMSK case (the dark line with rounded corners). Figure 11.15 shows that GMSK is a form of MSK. MSK exhibits a constant phase change during a symbol time, but the abrupt phase changes at symbol transitions cause MSK to occupy more spectrum than the equivalent rate GMSK. GMSK rounds off the phase transitions between symbols thus more tightly confining the spectral power to the allotted bandwidth. The smaller the BT product, the more power is confined to a portion of the spectrum by spreading the symbol changes over even more symbol
Introduction to GSM technology and implementation
441
GMSK
Phase
MSK
GMSK MSK
Data
time
Figure 11.15
MSK and GMSK.
periods. GMSK is said to have greater spectral compactness and efficiency than MSK, because GMSK confines the spectral emissions to a relatively narrow channel. The bandwidth-limited emissions of GMSK means we can use somewhat less power in the transmitter than we would if we used MSK. But the spreading of each information bit over three bit times, in the GSM case, adds ISI to the modulation even before it leaves the transmitter. So, the price we pay for the savings in spectrum and the power consumed in the transmitter by using GMSK, is picked up by the receiver, in which we are obliged to add even more complexity with an equalizer. 11.6.2
General equalizers An equalizer is any device that compensates for ISI in a receiver. Equalizers have been around for a long time. Simple impedance matching networks that perform echo canceling functions are examples of early devices still used today [39,40]. Figure 11.16 depicts a general form of a linear equalizer. Whenever we see an equalizer for the first time we should look for three parts: (1) the filter, (2) the control function, and (3) the feedback path.
442
GSM and Personal Communications Handbook
input d
a1
d
a2
d
a3
d
a4
a5
Σ
output
control algorithm
Figure 11.16
A linear equalizer.
The filter part in Figure 11.16 is a transmission line with taps separated from each other by a delay, d, of one symbol period, t, or d = t. As the distorted input from the receiver’s demodulator (see Figure 11.14) propagates down the filter, samples of the signal appear, at the filter’s sampling period, into the weighting functions, a. The outputs of the weighting functions are summed to form the equalizer output. The output will be a faithful reproduction of the original transmitted pulse if the weighting functions are adjusted properly. A look at the an levels in Figure 11.14 are an indication of what the weighting values should be. The weighting functions are adjusted by the control part in Figure 11.16 in such a way as to yield a flat frequency response with linear phase response. It can do this by amplifying the attenuated frequencies and attenuating the peaked ones. Such a filter is called a matched filter. The control part is usually some kind of algorithm contrived from a detailed knowledge of the channel it has to compensate for. If the channel characteristics do not change, such as in wire applications, then it is enough to hard wire a fixed control process. This makes the equalizer a fixed or static one. We can increase the resolution with which we define and compensate for the communications channel by oversampling the input, d < t.
Introduction to GSM technology and implementation
443
The feedback path in Figure 11.16 is a simple analog connection, which works fine for fixed and relatively well-behaved channels such as are found on phone lines. In mobile communications it is more common to use nonlinear feedback as is the case in the general adaptive device depicted in Figure 11.12. There is always some kind of decision device in digital systems between the demodulator and the decoder, which makes hard decisions on the state of symbols recovered in the demodulator. The decision device can be as simple as a comparator. Nonlinear feedback uses the decision device, which allows us to change the control algorithm in the equalizer with changing circumstances such as we would see in mobile radio channels. Such an adaptive equalizer is called a nonlinear feedback equalizer. Conceptually, the simplest forms of this kind of equalizer decode each symbol, or a group of symbols, adjust the equalizer’s actual output for the best agreement with the desired output, and then use the latest adjustments for the next round of symbols. There are several variations on the nonlinear technique. One of them is called the decision feedback equalizer (DFE), which inserts another matched filter, just like the forward one, in the feedback path coming back from the output of the decision circuit. The second filter in the feedback path is called the backward filter, and the gain outputs from its taps are summed with the forward filter’s gain outputs in the control block in Figure 11.16. This doubles the number of influences on the control algorithm; half of the inputs, the backward ones, are derived from the experience repairing the delay spread damage to the to previous time slot. Another nonlinear technique is called MLSE which is an elaborate and powerful technique that is covered in detail in Section 11.6.3. The MLSE technique can only work properly if it has some knowledge of what the desired data output from the decision device should be. It gets this information from a known symbol sequence that the transmitter sends to the receiver at regular intervals. The symbol sequences are called training sequences or sounding sequences. Though the appropriate symbol sequences do not look particularly simple to humans, they yield so-called ideal responses that only an equalizer likes and appreciates; they create signals that are relatively easy for the equalizer to examine, even through a poor channel, in order to figure out what the original, undistorted signal may have been. A bright green laser light is easier for us to distinguish through a dense fog among a riot of other lights and neon signs of various colors than a small white light would be. So, the symbol sequences are carefully
444
GSM and Personal Communications Handbook
constructed, through experiments or computer searches, to yield a simple output at the equalizer’s own filter. If the ideal response, in the absence of ISI, is simple enough, then the equalizer’s control algorithm can work backwards on a training sequence, which was received in the presence of ISI, to determine what the channel must have done to the sequence to account for the mess that was actually received. 11.6.3
Viterbi equalizer
A Viterbi equalizer includes a local signal generator and something called an increment calculator [27–29]. Viterbi equalization performs four daunting tasks. It constantly:
1. Generates its own versions of all possible data sequences that could come from the transmitter; 2. Calculates the receiver inputs that correspond to each and every transmitted possibility; 3. Compares the actual receiver inputs with the calculated ones; 4. Selects the locally generated data sequences that have the greatest probability of being the ones that were actually transmitted. The comparison task in step 3 is performed with metric calculations, which are the same as those used in convolutional decoders that use the Viterbi process. Because channel coding depends so much on convolutional codes these days, we will take the opportunity to expose the processes to the light of day in Sections 11.6.3.1 and 11.6.3.2. A generalized Viterbi equalizer is shown in Figure 11.17. At this point the reader may feel that the equalizer’s tasks dwarf all others in the GSM radio, or that, perhaps, something has been lost in the text. For now, we assure the reader that the tasks this kind of equalizer must perform would be, indeed, impossible were it not for the Viterbi algorithm (VA), which is the VA block in the figure, and which will be explained in due course. The Viterbi equalizer performs its work with vector quantities (complex numbers) and soft outputs that try to add some additional resolution to the digital symbol stream from the receiver’s demodulator. We will avoid obfuscation in our discussion by ignoring complex numbers and soft outputs as we confine ourselves to binary examples.
Introduction to GSM technology and implementation
445
Sequences input
Channel estimate
Signal generator
sounding Increment calculator
Demux
VA
information
Distortion
Figure 11.17
data output
Viterbi equalizer.
There are two paths in the Viterbi equalizer that terminate in the increment calculator just before the VA. The symbol stream from the demodulator enters a demultiplexer that separates the training sequence found in all the time slots and bursts (except in the FCCH) in the GSM air interface. The training sequence follows the upper path in Figure 11.17. The normal burst (NB) reserves 26 of its centermost symbols for one of eight time slot–specific training sequences, hence, the term midamble. We confine our attention in this discussion to the NB after we recall that the synchronization burst has 64 training bits and the access burst has 41 training bits. Remember, a symbol period is the same as a bit period in GSM. The bottom path in Figure 11.17 holds the remaining data symbols. An estimate of the channel characteristics is derived in the upper path. Take the simple case where a training sequence is extracted from a burst that arrived at the receiver over an ideal channel. If the training sequence is applied to a matched filter (Section 11.6.2), then a narrow impulse emerges from the filter. This is because the eight training sequences in the GSM time slots are carefully chosen to have this effect even when a great deal of delay spread occurs in the channel [27,29]. If the channel is not ideal, then delay spread will distort the pulse from the matched filter. Because the equalizer knows what the matched filter’s output should be for all eight of the training sequences, it can estimate the
446
GSM and Personal Communications Handbook
radio channel’s effects on the received signals by examining the output of the filter. This examination is also called a channel estimate. Such a channel estimate also renders a frequency offset value that can be used to synchronize (fine-tune) the receiver’s reference oscillator, and thus adjust it to the more stable frequency reference of the downlink or forward channel transmitted by the base station. Then, it adjusts the matched filter to get the proper response back again, and makes the same adjustments in the lower path of Figure 11.17 for the data that represent actual information. The signal generator in Figure 11.17 gets all possible transmitted data sequences from the sequences block above it. The local signal generator duplicates the distant transmitter’s modulator as it allows its output to be appropriately distorted by the current channel estimate. The bottom path has a filter that distorts the received data symbols the same way the locally created symbols are distorted in the modulator. The actual received data are compared with all the locally generated possibilities in an incremental calculator that yields confidence values, metrics. The VA uses the metrics to select the most likely data sequence that could have originated at the transmitter. The Viterbi equalizer differs from the other equalizers we looked at; it does not recover the received symbols, or even blocks of symbols, as linear techniques and the DFE does. Instead, the VA block in Figure 11.17 creates the most likely received symbol sequences itself and, hence, the data that probably came from the transmitter. Except for the input and output of the VA block in Figure 11.17, all the paths (the darker paths in the figure) carry I and Q values that describe the modulation from whatever source: (1) the recovered training sequences from the received time slots, (2) the received data, and (3) the output of the local signal generator. Moreover, in a practice reminiscent of the d < t oversampling technique described in Section 11.6.2, the symbol sequences described here are actually sequences of phase changes, which can be equated to symbol sequences. Finally, all the processes described here are performed numerically in a DSP after the demodulator’s baseband output is converted into appropriate digital values. 11.6.3.1
Convolutional encoder
Figure 11.18 depicts a small machine, which we will call a convolutional encoder, that illustrates the typical channel encoding process in digital transmitters [29,30,32]. The actual encoders used in GSM phones and
Introduction to GSM technology and implementation
447
m=3 n=2
+
information
1st 000101
0
0
0
001110001011
2nd +
encoded symbols
k=1
Branch
1
0
0
0
1
0
1
0
1
0
1
0
0
0
1
0
0
0 State
Status
Figure 11.18
Convolutional encoder.
base stations are much more complex than the simple machine depicted here; they have five stages (cells) in the shift register and are too complicated to serve as reasonable illustrations. The general principles are the same. The raw information input is on the left, and the channel coded output, sent out in pairs read left to right, is on the right. We will start with all zeros loaded into the input shift register on the left, and then shift the information k = 1 bit at a time into the shift register, which is m = 3 stages (cells) long. The status of the shift register is shown in the bottom of Figure 11.18 for all six input instances. The state of the encoder is the 2 bits already in the shift register, which will change with each new bit that appears in the branch cell, the darker cell on the far left of the shift register. The 2 bits stuck in the shift register (state) delay or hold the effect of the inputs on the left; we say that the encoder has a delay of 2. The encoder generates n = 2 bits output on the right for each single bit shifted into the
448
GSM and Personal Communications Handbook
left. The encoding rate, r, is written as r = k/n = ½; the output bit rate is twice the input rate. With each input instance, the entire shift register is regarded through two modulo-2 adders (which each are characterized by 0 + 0 = 0 and 1 + 1 = 0 and 0 + 1 = 1 and 1 + 0 = 1); the two-position switch on the output of the encoder passes through both of its positions. There is nothing particularly significant about the wiring between the modulo-2 adders, and the stages (cells) in the shift register in Figure 11.18; the machine serves only to illustrate several important baseband processes. Convolutional coders (CCs) and their codes are described thus: CC(n,k,m), which in this case is a CC(2,1,3) code. Figure 11.18 shows all 12 encoded outputs for the six inputs. Every 2 bits generated by the encoder is influenced by 3 consecutive bits on the input. The effect of the “smearing” of the encoder’s input over many output bits gives a properly equipped receiver the ability to correct errors in the received signal [41,42]. If n = 3, which would be a CC(3,1,3) code, then the receiver can make even more dramatic repairs to erroneous inputs, but the bit rate in the channel has to be increased to do so. The following paragraphs and figures illustrate how a receiver, with a detailed knowledge of how the transmitter’s encoder works, can correct errors in received data, and how these same techniques are used correct for ISI. 11.6.3.2
Convolutional decoder
A graphic way to describe undoing the convolutional code in Figure 11.18 is depicted in Figure 11.19, in which we accept the encoder’s undistorted and error-free output to see how a trellis diagram gets our original information back again [43]. The graphic rules we use to decode the bit stream is shown in the top of the figure. The four possible states of the encoder (in the two rightmost cells) appear on both sides of the small rule trellis in the top of Figure 11.19, and a state is represented by a dot. All eight possible transitions from one state to another state are indicated by a line over which an oval is drawn and the resulting two output symbols that are the result of the state change are written into the oval. The transition line can be dotted or solid depending on the branch bit as shown in the legend to the right of the rule trellis. Moving to the larger trellis in the bottom of Figure 11.19, the original information is represented by half the number of bits the encoder generated. The trellis diagram, so called because it resembles the framework of crossing strips of wood used to support climbing plants, is a tabulation of
Introduction to GSM technology and implementation 00
00
00
449
Transition of a pair of encoded symbols or a 1
caused by 0
11
10
in the BRANCH
10 00 11 10
01
01
01
01
11
11
10
00
00
2
00
2 11
1
2
3
2
00
0
0
11 11
10
0
0
0
11
0
0 10
0
00
0
10
0
01 0
01
0
2 2
11
t=0
Figure 11.19
1
2
3
4
5
6
Decoding rules and example.
all possible encoder state transitions at every symbol instant exactly as in the rule trellis [30,32]. We consider six transition instances starting with t = 1. As the receiver’s decoder recovers pairs of symbols, we plot all possible corresponding encoder states between the matrix of dots in Figure 11.19. Because there are four possible encoder states, there are four dots for each of the decoding instances, t. Because the encoder starts with all zeros, the decoder will begin the same way in the upper left corner of the figure at t = 0. As each pair of symbols is recovered in the receiver, at t = 1, for example, we draw a line from the stating dot (t = 0) to the next encoder state, which records the influence of the first information bit entering the machine on the left of Figure 11.18.
450
GSM and Personal Communications Handbook
There are 2k = 2 (with k = 1) possible branches from one state to the next, which is also the number of branches that could terminate in a dot (encoder state). A transition of encoder states caused by a “1” shifted into the encoder’s input is represented by a dotted line, and a solid line represents a state transition caused by a “0” shifted into the encoder’s input. Each pair of encoded output symbols from the encoder, which are recovered in the receiver, is shown in a small oval in the diagram in accordance with the rules in the top of the figure. At t = 1, we recover “11” and then calculate the branch metric, which is the confidence we have that the received pair of bits is, in fact, what was generated in the transmitter’s channel encoder. A transition from state “00” to “10” would mean the output symbols were “11,” which is no different from what we observed. 00 to 10 → 00 to 11 ⇒ 0 The symbolic shorthand contrived here is read: Encoder state “00” changed to the next encoder state “10” yields (→) encoder output pair “00” changed to subsequent encoder output pair “11.” The number of encoder output bits that would need to be inverted to give the observed output is (⇒) 0, which is the branch metric. The only other possibility is that the input to the encoder did not change (“00” to “00”), which would mean that the transmitted pair of symbols was “00.” This, however, would mean that we received both symbols in error (2 errors), which is less likely than the previous case. 00 to 00 → 00 to 00 ⇒ 2 We write the branch metric as a small number next to both of the possible states in the t = 1 column. Next, we write the current cumulative path metric as a larger number next to both states. The next pair of decoded symbols is “10,” which, if we also consider the second less likely case from the previous pair of decoded symbols, gives us four possibilities. We write the branch metrics next to each of the four possible states (dots) at t = 2. 00 to 00 → 10 to 00 ⇒ 1
Introduction to GSM technology and implementation
451
00 to 10 → 10 to 11 ⇒ 1 (which we exclude from the figure to avoid clutter) 10 to 01 → 10 to 10 ⇒ 0 10 to 11 → 10 to 01 ⇒ 2 But, at t = 2, one of the pairs of symbols already has a high path metric (less likely possible occurrence relative to the “00” state). If we calculate the four possible path metrics we get the following results, where the bold numbers indicate path metrics: 00 to 00 to 00 ⇒ 2 + 1 = 3 00 to 00 to 10 ⇒ 2 + 1 = 3 00 to 10 to 01 ⇒ 0 + 0 = 0 (survivor) 00 to 10 to 11 ⇒ 0 + 2 = 2 We write the path metrics in larger numbers next to all four possible dots at t = 2, and declare the path with the lowest path metric to be the survivor. We perform all of these tasks again for the next pair of symbols, “00” recovered at t = 3 as shown in Figure 11.19. Because trellis diagrams can quickly become too cluttered to be useful, we mark only a few of the branch and path metrics on the figure for t = 3 through t = 6. The convolutional decoder illustrated with the trellis diagram uses its input sequence to estimate the most likely stream of bits that caused the state transitions in the transmitter’s encoder. Because the example in Figure 11.19 has no errors in the recovered symbol stream, we choose the path with the lowest metric (0), which is the path with the highest confidence. Figure 11.19 distinguishes the most likely path from the less likely ones with bold lines. To determine the original information bits that entered the transmitter’s encoder, we simply note if a line between state transitions is dotted or solid. In Figure 11.19 the results are as follows: ◗ dotted line = 1; ◗ solid line = 0;
452
GSM and Personal Communications Handbook ◗ dotted line = 1; ◗ solid line = 0; ◗ solid line = 0.
This “10100…” sequence is exactly what we saw enter the channel encoder in Figure 11.18. The trellis diagram in Figure 11.20 illustrates how the same decoder can determine the original information sent into the transmitter’s convolutional encoder even when errors are present in the symbol pairs recovered in the receiver. To avoid calculating tedium and to keep the process clear, we change our transmitted data to a constant stream of zeros. In our example, the fourth bit recovered in the receiver is wrong: the “1” should have been a “0,” which means the second recovered pair of symbols is “10” instead of “00.” The process device between the encoder’s input information bits and the output symbol pairs in the top of Figure 11.20 represents whatever distorts the output stream, which is everything between
000000
Process
0
1
1
1
1
0
1
0
0
0
00
00
00
11
00
11
1
11
1
10
000000001000
00
00
1 0 00
3 2
2
2
2
10
2
10
0
1
01 01
4
01
2
2
1
2
3
11
t=0
Figure 11.20
1
4
5
Correct decoding in the presence of errors.
6
Introduction to GSM technology and implementation
453
the transmitter’s modulator and the receiver’s demodulator; the radio channel is the prime cause of errors. At t = 1, at which instant we received “00,” we calculate our branch metrics as before: 00 to 00 → 00 to 00 ⇒ 0 00 to 10 → 00 to 11 ⇒ 2 At t = 2 we recover the symbols “10” and continue with our branch and path metrics as if nothing were wrong. As before, the bold numbers are path metrics. 00000 to 00 to 00 ⇒ 0 + 1 = 1 00 to 00 to 10 ⇒ 0 + 1 = 1 00 to 10 to 01 ⇒ 2 + 0 = 2 00 to 10 to 11 ⇒ 2 + 2 = 4 We note that at t = 2, we cannot declare a surviving path, because we have two paths with identical path metrics, namely “1.” We will need to decode two more symbols when t = 3 before we can declare a survivor. We can also note that at t = 2, even as we have a branch metric of zero (at state “01”), it follows an unlikely event with a high branch metric of 2. At t = 3 we finally have a survivor (branch metric = 1) and all remains well through t = 6. Figure 11.21 illustrates what happens when the receiver is overwhelmed with symbol errors. In this example, we enlist the same stream of zeros into the transmitter’s channel encoder, but invert three out of the first four symbols at the receiver’s input. This makes the first pair of output bits “11” instead of “00” and the second pair “01” instead of “00.” Even if we look all the way out to t = 7, we see that the decoder interprets the encoder’s input incorrectly (1010101...) even if all the received symbols after t = 2 are received without error. One way to prevent this kind of decoding catastrophe is to add a little analog character to the decoder, which relieves the receiver’s decision function from making
454
GSM and Personal Communications Handbook
000000
2 2
00
00 11
000000000111
Process
4
3
2
2
4
4 2
00
5
0
2
00
0
1
2
11 0
0
11
11 0
3 0
10
1
10
00
1
2
1
2
10
10
00
1
3
00
3
01
4
1
1
01 01
1
01
01
3
2
1
1
11
1
01
1
10
t=0
Figure 11.21
1
2
3
4
4 1
10
5
6
7
Incorrect decoding in the presence of too many errors.
hard “yes” and “no” (“1” and “0”) decisions. The technique is called soft decision decoding. Soft decision decoding adds some additional resolution to the judgments that the receiver’s decision circuits make as symbols are recovered [29]. Figure 11.22 shows a simple case where a received symbol can be classified into one of eight categories, which could be something like a detected voltage in our simple example. It is not clear what the third symbol should be in the figure. A hard decision circuit may have declared the third symbol a zero (“0”), because there is an instant of high negative voltage at the beginning of the symbol time, or it may declare the third symbol a one (“1”), because the symbol, perhaps, spends more time in the “yes” or “1” territory relative to the “no” or “0” territory. If we take the same catastrophic situation depicted in Figure 11.21, but look closer and make the received symbol decisions soft ones, we can determine the correct transmitted information. In this case the surviving path metrics will have the greatest value rather than the value closest to zero. We will use the same scale in Figure 11.22 to resolve the received symbols as depicted in Figure 11.23. At t = 1 the branch metrics are
Introduction to GSM technology and implementation
455
4 3 2 1
time
−1 −2 −3 −4 1 Figure 11.22
0
?
1
Soft decisions.
00 to 00 → 00 to 00 ⇒ (−2) + (−1) = (−3) which reads: to go from 00 to 00 state, the signal has to be reduced (“–”) by a 2 and 1 metric, at their respective instances, to be in the negative domain. The (–3) reflects the probability, or our confidence, that the actually sampled values represent two 0s. 00 to 10 → 00 to 11 ⇒ 2 + 1 = 3 which reads: to go from 00 to 10 state (to render 11 at the imaginary encoder output), the signal has 2 and 1 “overshoot” or surplus metrics at their respective instances. There is a higher probability (+3), that the actual samples represent two 1’s. At t = 2 the branch metrics are 00 to 00 → 00 to 00 ⇒ 1 + 3 = 4 which means: to go from 00 to 00 state (to render 00 at the imaginary encoder output again), the signal overshoots, into the negative domain, by 1 and 3 metrics at their respective instances. The probability that the samples represent 0s is high.
456
GSM and Personal Communications Handbook
10 to 01 → 11 to 10 ⇒ (–1) + 3 = 2 similarly, to go from state 10 to 01, we have 1 negative metric too much for the first sample to go to 0 (–1) and we lack 3 metrics for a 1. 10 to 11 → 11 to 01 ⇒ (–1) + (–3) = (−4) Next, here we lack 1 metric to get a 1 and 3 metrics to get another one. Consecutively, at t = 3 the branch metrics are 00 to 00 → 00 to 00 ⇒ 4 + 4 = 8 01 to 00 → 10 to 11 ⇒ (–4) + (–4) = (–8) 01 to 10 → 10 to 00 ⇒ 4 + 4 = 8 At t = 4 and beyond, the received samples seem to have settled down in their representations of “0” symbols (–4) and the path metric climbs to 32 at t = 6 as the branch metrics settle to 8 on the surviving path. The absolute value of the level at any sampling instant shown at the top of Figure 11.23 is an indication of our confidence in the received symbol at that instant. The first sample is +2 and the second sample is +1. This means that we are a bit more sure that the symbol at the first sampling instant represents a “1” than we are of the sample at the second instant. Similarly, we are much more confident that the value sampled at the fourth instant represents a “0” than we are of the value sampled at the third instant. The first two recovered symbols, 2 and 1, respectively, are compared with both possible “clean” symbol representations, which are –4 and 4 on our high-resolution scale. We have almost no confidence the “2,1” pair of values declared during the first two sample instants actually represents 0,0 (–4,–4), so we assign a –3 branch metric: 00 to 00 → 00 to 00 ⇒ (–2) + –1)= –3 We have more confidence that the same two symbols, 2 and 1, actually represent 1,1. We, therefore, assign a higher metric: 00 to 10 → 00 to 11 ⇒ 2 + 1 =3
Introduction to GSM technology and implementation t=
1
3
2
457
5
4
6
+4 +3 +2
+1 Samples −1 −2 −3 −4
−3 −3
00
00
1
9
17
25
4
8
8
8
00
00
−8
11
10
00
00
33 8 00
−3
11
3
8
3 10
5
00
13
2
01 01
01
1
−2
13
13
0
11
10
t=0
Figure 11.23
1
2
3
4
13
0
0
10
5
6
Soft decision decoding.
If we calculate the branch metrics in this way out to t = 6, and then calculate the path metrics, we see that the path with the greatest metric gives the original transmitted data back to us. So, even as this discussion clarifies the channel coding and decoding blocks in the GSM transceiver, what does all of this have to do with equalizing the channel and working around the ISI problem? If we consider the GSM case with five stages (cells) in the encoder’s shift register, four stages of which constitute the status of the encoder, we turn back to Figure 11.17. At each sampling instant the received bit pattern, and one of 16 possible locally generated bit streams, is presented to the increment
458
GSM and Personal Communications Handbook
calculator as 32 inputs: 16 inputs for the received bit stream and another 16 for one of the local trials. The 16 local streams are constant until another time slot, with its training sequence, arrives to change the local signal generator’s output to reflect the latest channel conditions. The increment calculator issues 16 metrics to the VA block in Figure 11.17, one for each pair of inputs. In the manner demonstrated in Figure 11.23, the VA block generates the most likely data stream that was sent from the distant transmitter. Now, the symbols that are actually examined in a GSM receiver are not the simple soft values on a single scale seen in Figure 11.22. Instead, the instantaneous I and Q phase states are examined, and the dark lines of Figure 11.17 point out the signal paths that are actually complex numbers or I and Q values represented by complex numbers. The received signal’s phase state at any instant can be resolved into units of radians, or portions of a radian, in proportion to how closely we want to examine a recovered signal in order to make soft decisions on what the actual transmitted symbol probably was. But the use of complex numbers, and additional resolution in examining them, adds more complexity to the equalization process than Figure 11.23 may indicate. Some GSM-specific criteria for channel equalization can be found in [29] in which the reader will discover how easy it is to become overwhelmed in the details without computer simulations. Unlike the simple telephone line equalizers from which we have drawn liberal references in the chapter, the Viterbi equalizer hasn’t the faintest clue what the actual transmitted data may be; it merely knows how it was coded and the true nature of all the possible training sequences. It generates the most likely sequence of source data given the messy results from the receiver’s demodulator and its knowledge of the transmitter’s construction. The occasional training sequences help determine what is happening to the coded data on its way to the receiver. 11.6.3.3
Implementation
Typical implementations for the Viterbi equalizer, and the subsequent decoding functions, are done in a DSP. Some computations are suitably done by using so-called Viterbi accelerators, which are accessed by the DSP when carrying out the algorithms. For instance, the selection (by comparison) and update of soft path metrics, and the respective sequences, can be done in hardware whereas other functions like matched filter related calculations are realized in software. A GSM
Introduction to GSM technology and implementation
459
equalizer would use up to 16 states to recover the initially transmitted information. Less channel delay spread implies fewer states (perhaps only four states), which would demand less computational effort. Typical figures for equalizer implementations, including the help of accelerators, are in the range of 5 mips for the modem/equalizer, and one additional minute for the actual channel decoding (supported by a Viterbi accelerator). Other functions such as CRC calculations and burst assembly demand relatively little processing power.
11.7 Encryption and security in GSM GSM incorporates certain security features that prevent (1) unauthorized service access by maintaining the confidentiality of a subscriber’s identity and authentication on the SIM, and (2) over-the-air eavesdropping thanks to the ciphering of user and signaling data. The ciphering feature is optional. Subscriber anonymity is further protected by a number of features including the temporary mobile subscriber identity (TMSI), which, because it is a much shorter number than the IMSI, also frees up some signaling space on the air. References [44–50] provide relevant GSM specifications and other discussions on security and encryption. 11.7.1
Algorithms and keys
However the operators choose to use the security features included in GSM, the details and procedures behind them are thoroughly specified. Security mechanisms are realized with algorithms (A), keys (K), random numbers for challenges (RAND), and signed responses (SRES) to the challenges. Sensitive information, such as keys and algorithms, are never transmitted over the air. The mechanisms are transparent to the operators for interworking, and only sets of challenges and responses need to be exchanged between operators, or more precisely, between their networks on PCM links. Authentication keys and algorithms are stored in the SIM (see Chapter 8) on the mobile’s side of the channel, and within the authentication center on the network side. The SIM processes the 128-bit RAND challenge (a random number) from the network together with the 128-bit individual subscriber authentication key, Ki, as inputs to the authentication
460
GSM and Personal Communications Handbook
algorithm, A3. The 32-bit result is called the signed responses, SRES, which the MS sends back to the network. The network, which has a table of current Ki assignments listed by subscriber identity (in the VLR and/or HLR), compares the mobile’s SRES with its own calculations to determine if the subscriber is, in fact, who he says he is before granting access to the system. Refer to Chapter 8 for graphical explanations. A similar approach is taken with ciphering. The ciphering algorithms are implemented in the mobile equipment (ME) and in the network (BTS). Because they need to run in real time during a phone call, their implementation needs to exhibit good performance. Typically, an encryption engine is built in silicon hardware rather than in DSP software, and consists of only a few thousand transistors [20]. One parameter, the cipher key, Kc, is used in the encryption engine, A5, to cipher the coded transmitter data and perform the corresponding decoding process in the receiver. The Kc value is computed in a manner almost identical to the way the SRES value is computed during authentication. The SIM, which holds the A8 cipher key algorithm, accepts the same RAND challenge the network passed to the mobile (usually on a SDCCH) during authentication, and combines it with Ki within A8 to deliver Kc. Additional security is evident by noting that the SIM does not reveal Ki or the A3 and A8 algorithms, which can be specific to an operator. Key Kc is derived from Ki and RAND, which means that Kc can change during a conversation, and will certainly be different with each call setup. 11.7.2
Ciphering in GSM
GSM ciphering algorithms are applied to protect user and signaling data immediately before they are sent over the air. The protection is accomplished by manipulating the 114 radio bits that fill each normal burst in a TDMA frame. A reverse manipulation is performed in the receiver immediately after the data stream is recovered in the receiver. A ciphering sequence and a deciphering sequence have to be generated for each TDMA frame by a stream cipher algorithm. A stream cipher algorithm with so-called linear feedback shift registers (LFSRs) is used in GSM. The A5 is first fed with the 64-bit cipher key Kc (for initialization) and the current 22-bit TDMA frame number. Note that Kc may actually contain less than 64 bits of significance; 64 significant bits implies maximum security. A short (not-so-secure) Kc is accompanied by enough zeros in order to fill all 64 bits of the A5 register. Because the output is also a function of the
Introduction to GSM technology and implementation
461
current TDMA frame number, the output of the clocked A5 shift registers, the ciphering sequences, is different for each TDMA frame and has two different uplink and downlink cipher sequences of 114 bits each. Finally, the payload data (received bursts and transmitted bursts), each of which holds 114 bits, is XORed with the current ciphering sequence. To recover the transmitted payload data, the receiver needs to use the same key for decryption (XOR with received radio bits) that was used by the transmitter for encryption. 11.7.3
Regulations
The A5 encryption algorithm, developed in Britain, was originally intended to be used in the Pan-European cellular system for which the GSM specifications were initially defined. Its use was confined to CEPT countries (Western Europe). Since the A5 algorithm represents a sophisticated ciphering technology, there were valid concerns from Western governments about protecting it and preventing its misuse. Tight export regulations (for export from Great Britain) are applied with nondisclosure obligations that must be observed by the manufacturing industry, which receives the technical details of the A5 algorithm. To address the need to export GSM technology to countries outside Western Europe, a “weaker” A5 algorithm, dubbed the A5/2 algorithm, was drafted and adopted. The original A5 was subsequently renamed the A5/1 algorithm. This leaves GSM networks with three options for payload data protection: A5/1, A5/2, or no ciphering at all. One has to look where (in which country) a network operates, and then look further at the network operators’ choices to sort out the ciphering possibilities. Export restrictions on A5/1 and A5/2 implemented in silicon within mobile terminal equipment are not particularly strict. Today’s GSM terminals are usually fitted with both algorithms. The GSM specifications allow the use of up to seven different A5s. At the time of this writing, no further A5 algorithms have been specified beyond the A5/2. The administration of the GSM encryption secrets around the A5/1 and A5/2 algorithms are left with the GSM MoU’s security group. 11.7.4
Security vs. fraud
Together with the SIM, a GSM terminal can provide a very high level of security for both the equipment and the subscriber. We have all heard the
462
GSM and Personal Communications Handbook
reports on how revealing authentication keys and cracking ciphering keys impairs GSM security, but the technical effort for doing so renders the tasks fruitless. We have also heard the reports about simple FM scanners affording easy and low-cost eavesdropping on some analog cellular and cordless systems. With the introduction of digital technology, the technical effort for over-the-air snooping has increased dramatically. The pure digital nature of the information transmitted over the radio channel, and the fact that digital speech coding is used together with channel coding, further raises the price of unauthorized listening sessions. Moreover, the TDMA effect in GSM and IS-136 systems, the signal spreading used in CDMA systems, and frequency hopping and digital encryption techniques move the cost of casual radio interceptions beyond what most people who engage in such activity can afford. Protecting digital information is relatively simple and inexpensive, and it can be very effective. In particular, the GSM ciphering algorithms, and the related radio techniques, can be regarded as extremely safe barriers against determined interceptors. Some safety considerations, crime prevention efforts, and criminal investigations ask for particular police and law enforcement agencies to be able to listen into phone conversations. This can be done quite easily through appropriate technology in the fixed backbone network in which we find access points where encryption is not applied to the signaling and user data, and the difficulties of dealing with the TDMA effect are absent. Finally, we are all aware of the techniques employed in cloning cellular phones through, for example, copying electronic serial numbers (ESNs), which in some analog cellular systems is sufficient to place calls at a legitimate subscriber’s expense. Again, digital technology on the air interface and the introduction of sophisticated authentication mechanisms help to prevent this kind of fraud. Progress has been made in network management software, which alerts the operator and bars the service for a particular subscription when, for instance, calls are simultaneously made from what seems to be the same terminal in different locations, or when too short a time passes for a legitimate terminal to travel between two locations.
11.8
Mixed signals
Digital telecommunications technology cannot work without access to the analog domain. Human voice originates and terminates with analog
Introduction to GSM technology and implementation
463
signals. Today, even though we can reach multigigabit per second transmission speeds, the physical transmission medium remains an analog one. Signals need to be converted from the familiar analog domain into the digital domain in which computers, processors, and control logic perform their work so well, and then back to the analog domain again. As we saw in our block diagrams of the different signal processing paths in a GSM mobile station, there are three two-way interfaces: (1) the voice codec, (2) the baseband codec, and (3) the radio control interface. Silicon implementations that combine all three functions are common in analog complementary metal-oxide semiconductor (CMOS) processes, which are not suited for adding too many digital functions due to size, cost, yield, and power consumption. Today we can even see mixed signal technology realized with “digital” CMOS processes. This implies the integration of “analog” functions (converters) and cells with all the additional digital baseband logic on one piece of silicon. Typical approaches primarily use the “digital” sigma-delta (∑∆) concept for conversions in interfaces 1 and 2 just mentioned. This means that analog signals are modulated with a high sampling rate that yields ones (“1”) for increasing analog signals (the voltage of the new sample is higher than the previous sample) and zeros (“0”) for decreasing signals (the voltage of the new sample is lower than previous sample). A digital decimation filter delivers PCM words at the required resolution and rate. For a 10-bit A/D converter in the GSM RX path (at 270,833 Ksps), using an 8× oversampling rate, the minimum sampling/modulation rate is 10 × 270,833 × 103 × 8 = 21.666 MHz. This is almost the double rate of the 13-MHz GSM reference clock that is used for this purpose (26 MHz), thus leaving some overhead for serialization and deserialization. Advantages of the ∑∆ concept are the suppression of quantization noise (through shifting the noise spectrum above half the sampling frequency), and the fact that further filtering can be performed digitally. Resolution and accuracy depend on the order and sampling rate of the ∑∆ modulator and the low-pass, postfilter characteristics. Depending on which technology is available, analog filtering (for anti-aliasing or reconstruction) can be supported on-chip with capacitors and resistors. By this, interfacing serialized data with other components, which has its own power consumption penalties and spurious signal generation tendencies, is not necessary. A reduced component and pin count on the printed circuit board allows for greater and more cost-effective integration and
464
GSM and Personal Communications Handbook
manufacturing. Some external passive and active components are required when on-chip technology is too big or inaccurate. A ∑∆ modulator can be used in the receiving path of the voice codec for D/A conversion that reconstructs the analog voice signal from a digital bit stream received from the speech codec through an interpolation filter. D/A conversion and filtering, for example, in transmit path 2, has to deliver a band-limited analog signal from a digital data stream. The signal shaping (modulation), which is GMSK with B × T = 0.3, is typically done through a look-up table that holds symbol transition curves. If, for example, a particular GSM realization selected an 8× oversampling rate, the modulated data stream would have a rate of 8 × 270,833 Ksps, which is represented by an input signal to the D/A converter (typically of the switch capacitor type in CMOS implementations) with a 2.1667-MHz (13/6-MHz) rate. Other RF functions (interface 3) include D/A converters for: ◗ Automatic frequency control (AFC), which is the tuning reference of
a VCO; ◗ Automatic gain control (AGC), which is used in LNA stages; ◗ Power amplifier control (PA), which performs the ramping and con-
trol of RF transmitter output power levels. The requirements for speed and resolution are typically in the range of several microseconds and a resolution of 10 bits is typical. Additional A/D converters are called on for: ◗ Received signal strength measurements (for RF carrier acquisition
and monitoring); ◗ Battery voltage and supervision; ◗ Temperature supervision in PA stages.
Again, the required conversion time is slow (a few microseconds) and a resolution of 6 to 10 bits is adequate to serve the purposes. A single A/D converter can be shared among different functions, because not all of them need to be done simultaneously and constantly. This implies the use of a signal path multiplexer.
Introduction to GSM technology and implementation
465
Other mixed signal functions include a system PLL synthesizer, which generates all of the required system clocks off the reference oscillator frequency, which is 13 MHz in GSM.
11.9 Microprocessor control As we can see when we look back at Figure 11.6, the heart of the GSM mobile is a control processor (near the center of the figure). The mobile equipment in the GSM networks was assigned many control and management functions. The mobile terminal is a miniature counterpart to the whole GSM network. It has to support the network’s connection management, mobility management, and radio resource management tasks, sundry user communication protocols (data and voice), and a more or less sophisticated user interface. The higher level intelligence above the complex real-time digital signal processing layers are built into the GSM mobile station as software running on the microprocessor. The microprocessor is quite often a reduced instruction set computer (RISC) with a 16bit instruction set and external (off-chip) memories for programs and data (flash ROM and RAM). Usually a real-time operating system (RTOS) supports preemptive and interrupt-based software design. A TDMA timer (see Section 11.10) furnishes interrupts to the microprocessor. Highpriority control tasks are executed before lower priority protocol and user interface tasks. In general, the microprocessor treats lower layer functions implemented in the DSP, interfaces, and other proprietary logic as black boxes. Information is polled as it is needed, and instructions and parameters are exchanged through commands in shared memory or registers. The following high-level tasks are typically covered by the microprocessor: ◗ User interface and application software (menus, keypad/display
interactions, and access to services above layer 3); ◗ Protocol (GSM layer 1; logical channel handling; frame multiplex-
ing for hyperframes, superframes, and the multiframes; time and frequency tracking; and the layer 2 and layer 3 functions);
466
GSM and Personal Communications Handbook ◗ Interface to, and communications with, the SIM; ◗ Interface to the digital signal processor; ◗ Interfaces to hardware (e.g., the radio for control functions); ◗ Interface to the serial data port for fax and data services.
A typical GSM implementation requires relatively low processing power, compared to the DSP, and is generally in the range of several million instructions per second and several hundreds of kilobytes of program ROM. The user interface (UI) software occupies a huge chunk of the program and data memories. The more complex, sophisticated, elaborate, and the more services that are supported, the larger the UI software effort and code size. UI design is an art of its own and is a very important one. The user interface, its ease of use, and its general look and feel are some of the key differentiating factors among manufacturers apart from size, price, weight, and standby time. It also represents the aspect of current GSM phones that needs the most work; ask any owner of a GSM phone how to activate call divert, how to retrieve his short messages, or how to retrieve a stored phone number from her phone, and you are likely to wait a long time for the correct reply.
11.10
GSM timing
All GSM terminal timing references are derived from a 13-MHz clock. The oscillator does not have very stringent free running stability requirements (1 to 3 ppm), because exact synchronization (0.1 ppm) is derived from the downlink signal from the GSM base station (see lbelow). The baseband bit rate is 270,833 kHz/Kbps, which is 13 MHz divided by 48, thus one symbol lasts 3.69 µs. One TDMA frame is 4.615 ms, which is eight time slots. Every time a 13-MHz counter runs up to 60,000, a new TDMA frame begins. Thus, a 26 multiframe (see Figure 11.24)—consisting of 24 frames for TCH, 1 for SACCH (S) , and 1 left idle (I)—lasts 120 ms [2,51]. Two traffic channel multiframes consist of 52 basic TDMA frames. This is one TDMA frame more than makes up the 51 multiframe, which is used in all kinds of signaling channel combinations including, for example, the base station control channel, which is a combination of FCCH, SCCH, BCCH, and CCCH resources. This 51–26 relation allows a mobile station
Introduction to GSM technology and implementation 0
1
2
3
4
5
6
467
7
1 TDMA frame = 8 time slots 2*26-multiframes = 2*26 = 52 TDMA frames ... 0 1 2 3 4 51-multiframe = 51 TDMA frames .. 0 1 2 3 4
Figure 11.24
12
I 22 23 24 25 0 1 2 3 4
S 22 23 24 25 26 27 28 29 30
12
I 22 23 24 25 ...
S 48 49 50 0 1 2 3 4 5
...
The 26- and 51-multiframe structures.
engaged in a traffic channel to monitor the base channel as the idle frames of the 26 structure slowly slide over the 51 structure of the base channel. In Figure 11.24 the TCH idle frame (I) occurs at signaling frames 30 and 5 (51-multiframe structure). The next two TCH idle frames would fall on signaling frames 31 and 6, respectively. So the MS can sequentially grasp all the necessary information (synchronization, cell/system information) of the serving base station as well as that from adjacent cells. The lower level logical timing in GSM base stations and mobile stations is typically and practically performed on a per-TDMA-frame basis. The events include setting radio parameters for receiving, transmitting, or monitoring functions (channel number/frequency, channel parameters for PLL settings, power, low noise amplifiers, etc.) as well as logical information. Do we need to generate a traffic channel burst or a SACCH, or do we need to send a SDCCH? The logical channels need to be scheduled. The use of interrupt controlled preemptive operating systems, for both the controller and the DSP, assists in the flexible scheduling of different tasks according to predefined priorities. Still, because timing in GSM systems may shift or drift (through frequency offsets or a time delay when moving), and correction mechanisms are in place (timing advance commands), a higher resolution than the single-bit period (e.g., 1/4 or 1/8 bit periods) is called on for precise adjustments. A GSM base station derives its highly accurate absolute timing reference off the digital Abis network interface clock. This clock is locked to a rubidium reference somewhere in the nodes connected with PCM links. A GSM mobile station locks onto its master, which is the BTS. Coarse timing and frequency correction is achieved through initial coarse synchronization with the FCCH and then with the SCH. Once locked onto the
468
GSM and Personal Communications Handbook
base channel, a mobile station maintains fine synchronization through its tracking of the paging channel as the equalizer provides sufficient frequency error information off the downlink training sequences for reference clock adjustments. Further coarse timing correction is carried out through timing advance commands, which are calculated and triggered by the BTS [2].
11.11 Components and technology Advances in digital technology let us put more and more baseband processing into chips, thus reducing component count, increasing reliability, diminishing the size of handsets, and lowing power requirements. Intense processing functions, such as voice coding, microprocessor control (protocol stack, user interface), signal processing (modem), RF/IF stages, and mixed signal functions used to be integrated on separate components. This was due to the lack of highly integrated components (RISC microprocessor and DSP). CMOS silicon technology (above 0.6 µm) did not allow integration of multiple entities onto one or two chips. Today’s technology and transistor density, which is typically 0.35- and 0.25-µm technology, allows comprehensive integration of GSM processing functions that incorporate all of the digital baseband entities: ◗ RISC microprocessor; ◗ Digital signal processor (DSP); ◗ Memory (for DSP and microprocessor cache), except flash memory
that holds microprocessor code and data, as well as related microprocessor RAM; ◗ Supporting logic (timer, registers, accelerators, encryption engines,
etc.). When silicon process technology allows the integration of mixed signal blocks with sufficient suppression of substrate noise and other prerequisites, then all the required baseband functions can be integrated onto one piece of silicon. This is the trend for GSM baseband chips, ASICs, ASSPs, and chip sets. Integrating mixed signal logic with the digital logic
Introduction to GSM technology and implementation
469
does away with the need to drive serialized data signals between components over printed circuit boards, thus saving power. Lots of progress has been made in the RF sections even though the evolution of the technology has not come as fast as it has in digital CMOS technology. Initially, the RF circuits were built from standard discrete components. With the market drive behind digital cellular technology, RF and IF integrated circuits (ICs), bipolar silicon technology, BiCMOS, and gallium arsenide (GaAS) monolithic microwave integrated circuits (MMICs) have appeared that allow small size and low-power RF front-ends and IF implementations [52]. Shrinkage of silicon technology and related silicon process enhancements allow a reduction of the operation voltages. Earlier solutions depending on a 5V supply for both baseband and IF/RF circuits can be substituted for newer technologies that run off a nominal 3V supply (±10%). Power requirements for CMOS baseband ICs are moving from 2V down to 1V. The RF power amplifiers still require higher voltages. Trade-offs occur between the desired RF power output (in GSM handheld applications this can be up to 2W), supply and breakdown voltages, and impedance matching [53,54]. Typical figures for average GSM power consumption (off a 3V supply) are in the range of a few milliamps, where 5 mA is a typical example for standby operation (listening to a paging channel and monitoring adjacent cells) and several hundreds of milliamps (150 mA is typical) for the conversation mode. These figures depend on many parameters and only typical values can be given. For example, the periodic appearance of the paging channel for a particular serving cell is variable, and can triple the best case figure for the worst case for battery life between charges. The organization of paging messages is intermittent with up to a few seconds between phone-specific pages. Phones need to monitor periodically the serving cell and adjacent cell signals, and this affects the duty cycles during which a mobile phone needs to activate some sections of its circuitry for receiving and processing the information. Transmission in standby mode is normally not required. There can be intermittent short periods when a mobile phone needs to leave the idle status and communicate (transmit) with the network: location updating/registration with the network when roaming. A GSM terminal may have a (baseband) duty cycle for standby activity (reading the paging channel and full monitoring) of as little as 0.5%!
470
GSM and Personal Communications Handbook
Designs can exploit sleep modes. The processor, baseband functions, and the radio blocks can be turned off during the idle periods. Even the whole reference oscillator section (13 MHz) can be switched off during such periods; in such case it can be substituted by a low-frequency, lowpower oscillator operating at 32 kHz. Conversation or talk time largely depends on the output RF power control levels, and the use of voice activity detection and discontinuous transmission. The RF power amplifier in conversation mode uses up most of the energy, whereas in standby mode the baseband section has at least an equal impact on the available energy compared to the whole radio. This means that a battery with a typical 500 mA-hour (mAh) capacity will support about 100 hours of standby and several hours of talk time.
11.12 Guide to the literature The following paragraphs organize some of the references in this chapter so that they can be pursued in an order anyone new to digital radio will find helpful. They are listed in the order in which they should be read. Some of the books are old and may be hard to find, but the effort will be rewarded. The references listed here should not be considered the result of an exhaustive search of everything available; they are merely what the authors have found particularly helpful in their own work. 11.12.1
General radio design Bowick’s work [8] is a small workbook that is an excellent introduction to RF circuits. Vizmuller [9] provides a more modern treatment of the same subject, and his book is an excellent introduction to the design process. It deals with the passive components and circuit details very well, and is helpful in sorting through all the compromises in RF design. Larson [10] gives many excellent examples of digital techniques, and offers clear and thorough explanations of modulation and demodulation. 11.12.2 Coding and its mathematics
Pierce [42] provides an introduction to information theory, and Tarasov [41] introduces probability; both are aimed at the general reader. Wiggert
Introduction to GSM technology and implementation
471
[43] followed immediately by Peterson [35] take the mystery out of error codes. The Berlekamp publication [34] is a collection of important papers, some rather old, on coding theory written by key figures in the field. 11.12.3
Digital radio
Aside from such classics as Taub and Schilling’s Principles of Communications Systems (McGraw-Hill), and Proakis [33], which should be in every engineer’s private library; the following are worth looking for. Start with [39], which, though Bellamy deals mostly with wireline systems, is a simple and clear introduction to digital communications. Follow this with Rappaport’s book [28], which is a relatively short but complete survey of wireless practice. Feher [32] concentrates on digital radio, and his book has beautifully clear drawings and illustrations. Finish off with Steele [26], who provides a collection of separate works from several authors that deal exclusively with digital cellular radio. There is a separate chapter at the end that uses GSM as a review of everything in the book. Once the reader is familiar with digital radio, four more books will fill in the empty spots in GSM. Start with [36] as an excellent survey of cellular radio. Follow this with the Redl et al. book [2], which is, as its title says, an easy introduction to GSM. Finish up with a new book by Mehrota [27], which is an amplification and elaboration of the material in [2], but without the material on testing.
References [1]
GSM TS 06.10 (ETS 300 580-x), “Digital Cellular Telecommunications System, GSM Full Rate Speech Transcoding,” ETSI, Sophia Antipolis.
[2]
Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 5.
[3]
GSM TS 05.03 (ETS 300 575), “Digital Cellular Telecommunications System, GSM Channel Coding,” ETSI, Sophia Antipolis.
[4]
GSM TS 06.20 and related documents (ETS 300 581/1–8), “Digital Cellular Telecommunications System, Half Rate Speech Transcoding,” ETSI, Sophia Antipolis.
[5]
GSM TS 06.51 and related documents (ETS 300 723–730), “Digital Cellular Telecommunications System, Enhanced Full Rate Speech Transcoding,” ETSI, Sophia Antipolis.
472
GSM and Personal Communications Handbook
[6] GSM TS 05.04 (ETS 300 576), “Digital Cellular Telecommunications System, Modulation,” ETSI, Sophia Antipolis. [7] Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 10. [8] Bowick, C., RF Circuit Design, Indianapolis, IN: Howard W. Sams & Company, 1982, Chap. 7. [9] Vizmuller, P., RF Design Guide Systems, Circuits, and Equations, Norwood, MA: Artech House, 1995, Chap. 1. [10] Larson, L. E. (ed.), RF and Microwave Circuit Design for Wireless Communications, Norwood, MA: Artech House, 1996, Chap. 4. [11] Redl, S. M., M. K. Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 10. [12] McMahan, Khatzibadeh, and Shah, Wireless Systems and Technology Overview, Dallas, TX: Texas Instruments, (www.ti.com), 1996. [13] Maoz, B., and A. Adar, “GaAs IC Receivers for Wireless Communications,” Microwave Journal, Vol. 39, No. 1, January 1996. [14] Winter, E., “Speech Coding,” Communications Systems Design Magazine, May 1996. [15] Xydeas, C. “Speech Coding,” in Personal and Mobile Radio Systems, R. C. V. Macario (ed.), IEE Telecommunications Series 25, London: Peter Peregrinus, 1991. [16] http://www.speech.su.oz.au/comp.speech/ and http://www-mobile.ecs.soton.ac.uk/speech_codecs and http://wwwdsp.ucd.ie/speech/tutorial/speech_coding/. [17] Steele, R., “Speech Codecs for Personal Communications,” IEEE Communications Magazine, November 1993. [18] Rappaport, T. S., Wireless Communications Principles & Practice, Upper Saddle River, NJ: Prentice Hall PTR, 1996, Chap. 7. [19] Qualcomm Data Sheet for the Q5414 Variable Rate Vocoder, San Diego, CA: Qualcomm ASIC Products. [20] Steele, R., et al., Mobile Radio Communications, London: Pentech Press, 1992, Chap. 3. [21] GSM TS 11.10 (ETS 300 607), “Digital Cellular Telecommunications System, GSM Mobile Station Conformity (Test) Specifications,” ETSI, Sophia Antipolis. [22] GSM TS 03.50, (ETS 300 540), “Digital Cellular Telecommunications System, Transmission Planning Aspects of the Speech Service in a GSM PLMN,” ETSI, Sophia Antipolis.
Introduction to GSM technology and implementation
473
[23] GSM TS 06.10 and related documents (ETS 300 580-x), “Digital Cellular Telecommunications System, Full Rate Speech Transcoding,” ETSI, Sophia Antipolis. [24] Spanias, A., “Speech Coding: A Tutorial Overview,” IEEE Proceedings, Vol. 82, No. 10, October 1994. [25] Vary, P., “GSM Speech Codec,” Digital Cellular Radio Conf. Proc., Hagen, Germany: Deutsche Bundespost, France Télécom, Fern Universitaet, 1988. [26] Steele, R., et al., Mobile Radio Communications, London: Pentech Press, 1992, Chap. 4. [27] Mehrotra, A., GSM System Engineering, Norwood, MA: Artech House, 1997, Chap. 6. [28] Rappaport, T. S., Wireless Communications Principles & Practice, Upper Saddle River, NJ: Prentice Hall PTR, 1996, Chap. 6. [29] Steele, R., et al., Mobile Radio Communications, London: Pentech Press, 1992, Chap. 8. [30] Yacoub, M. D., Foundations of Mobile Radio Engineering, Boca Raton, FL: CRC Press, 1993, App. 6A. [31] Gibson, J. D. (ed.), The Mobile Communications Handbook, Boca Raton, FL: CRC Press, 1996, Chap. 5. [32] Feher, K., Wireless Digital Communications Modulation & Spread Spectrum Applications, Upper Saddle River, NJ: Prentice Hall PTR, 1995, Chap. 5. [33] Proakis, J. G., Digital Communications, 2nd ed., New York: McGraw-Hill, 1989, Chap. 2. [34] Berlekamp, E. R. (ed.), Key Papers in the Development of Coding Theory, New York: IEEE Press, 1974. [35] Peterson, W. W., and E. J. Weldon, Jr., Error Correcting Codes, 2nd ed., Cambridge, MA: The MIT Press, 1990. [36] Balston, D. M., et al., Cellular Radio Systems, Norwood, MA: Artech House, 1993, Chap. 7. [37] Viterbi, A. J., CDMA Principles of Spread Spectrum Communication, Reading, MA: Addison-Wesley, 1995, Chap. 4. [38] Redl, S. M., M. K.Weber, and M. W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 13. [39] Bellamy, J., Digital Telephony, 2nd ed., New York: John Wiley & Sons, 1991, Chap. 1. [40] Ericsson home page, Echo in Telephony Systems, http://www.ericsson.com/echo/edu/we.htm.
474
GSM and Personal Communications Handbook
[41] Tarasov, L., The World Is Built on Probability, Moscow: Mir Publishers, 1988, Chap. 3. [42] Pierce, J. R., An Introduction to Information Theory, New York: Dover Publications, 1961, 1980, Chap. 8. [43] Wiggert, D., Codes for Error Control and Synchronization, Norwood, MA: Artech House, 1988, Chap. 6. [44] Redl, S. M., M. K. Weber, and M.W. Oliphant, An Introduction to GSM, Norwood, MA: Artech House, 1995, Chap. 3. [45] GSM TS 02.09 (ETS 300 506), “Digital Cellular Telecommunications System, Security Aspects,” ETSI, Sophia Antipolis. [46] GSM TS 02.17 (ETS 300 509), “Digital Cellular Telecommunications System, Subscriber Identity Modules,” ETSI, Sophia Antipolis. [47] GSM TS 03.20 (ETS 300 534), “Digital Cellular Telecommunications System, Security Related Network Functions,” ETSI, Sophia Antipolis. [48] Vedder, K., “GSM Security, Services and the SIM—Developments in Technology in the Fight Against Fraud,” GSM ’96 Congress, Cannes, France, Feb. 20–22, 1996. London: IBC Technical Services. [49] Brookson, C., “Has GSM’s Security Been Compromised?” GSM ’96 Congress, Cannes, France, Feb. 20–22, 1996. London: IBC Technical Services. [50] Schneier, B., Applied Cryptography, New York: John Wiley & Sons, 1994. [51] GSM TS 05.02 (ETS 300 574), “Digital Cellular Telecommunications System, Multiplexing and Multiple Access on the Radio Path,” ETSI, Sophia Antipolis. [52] Abidi, A. A., “Low Power Radio Frequency ICs for Portable Communications,” Chap. 3 in RF and Microwave Circuit Design for Wireless Communications, L. E. Larson (ed.), Norwood, MA: Artech House, 1996. [53] Baringer, C., and C. Hull, “Amplifiers for Wireless Communications,” Chap. 79in RF and Microwave Circuit Design for Wireless Communications, L. E. Larson (ed.), Norwood, MA: Artech House, 1996. [54] Vizmuller, P., RF Design Guide, Norwood, MA: Artech House, 1995.
Appendix Coding of the default GSM alphabet
E
ach character is represented by a pattern of 7 bits (b1 ... b7). These characters are packed into the user field of a short message according to Sections 6.1.2 and 6.1.3.
475
476
GSM and Personal Communications Handbook
b7
0
0
0
0
1
1
1
1
b6
0
0
1
1
0
0
1
1
b5
0
1
0
1
0
1
0
1
0
1
2
3
4
5
6
7
b4
b3
b2
b1
0
0
0
0
0
SP
0
¡
P
¿
p
0
0
0
1
1
£
1)
!
1
A
Q
a
q
0
0
1
0
2
$
F
“
2
B
R
b
r
0
0
1
1
3
¥
G
#
3
C
S
c
s
0
1
0
0
4
è
L
¤
4
D
T
d
t
0
1
0
1
5
é
W
%
5
E
U
e
u
0
1
1
0
6
ù
P
&
6
F
V
f
v
0
1
1
1
7
ì
Y
‘
7
G
W
g
w
1
0
0
0
8
ò
S
(
8
H
X
h
x
1
0
0
1
9
Ç
Q
)
9
I
Y
i
y
1
0
1
0
10
LF
X
*
:
J
Z
j
z
1
0
1
1
11
Ø
1)
+
;
K
Ä
k
ä
1
1
0
0
12
ø
Æ
,
cell L Ö
l
ö
1
1
0
1
13
CR
æ
-
=
M
Ñ
m
ñ
1
1
1
0
14
Å
ß
.
N
Ü
n
ü
1
1
1
1
15
å
É
/
O
§
o
?
Glossary
Glossary µs
Microsecond, 1/1,000,000 of a second
∑∆
Sigma-delta
3GIG Third-generation Interest Group, a technology group within the GSM MoU looking after third-generation systems A/D
Analog-to-digital conversion
AAeM
Automatic answer of eMLPP service, supplementary service
ABM
Asynchronous balanced mode, state of the radio link protocol
AbS
Analysis by synthesis
ACC
Access control class, elementary file on the SIM
ACELP ACM
Algebraic code(-book) excited linear predictive Accumulated call meter, elementary file on the SIM
ACMmax
ACM maximum value, elementary file on the SIM
ACTE Advisory Committee for Terminal Equipment (type approval committee of EEC) ACTS AD
Advanced Communications Technologies and Services Administrative data, elementary file on the SIM 477
478
GSM and Personal Communications Handbook
ADC
Analog-to-digital converter
ADM Asynchronous disconnected mode, state of the radio link protocol ADN Abbreviated dialing number, elementary file on the SIM containing the phone book ADPCM AF
Adaptive differential pulse code modulation
Audiofrequency
AFC
Automatic frequency control
AGC
Automatic gain control
AGCH AIG
Access grant channel Arab Interest Group, an interest group within the GSM MoU
AIReach
Hughes Network Systems’ version of PACS
ALS Alternate line service, PCN-defined service that allocates two phone numbers (lines) to one subscriber AM AMPS
Amplitude modulation Advanced mobile telephone system
AMR
Adaptive multirate (speech codec)
ANSI
American National Standards Institute
AoC AoCC AoCI
Advice of charge, supplementary service (SS) Advice of charge charging, subgroup of the AoC SS Advice of charge information, subgroup of the AoC SS
APCO Association International APIG MoU ARCH ARQ
of
Public-Safety
Communications
Officials
Asia Pacific Interest Group, an interest group within the GSM Access response channel (IS-136) Automatic repeat request
Glossary
479
ASIC
Application-specific integrated circuit
ASSP
Application-specific standard poduct
BABT
British Approvals Board for Telecommunications (UK)
BAIC
Barring of all incoming calls, subgroup of call barring SS
BAOIC ring SS
Barring of outgoing international calls, subgroup of call bar-
BAOIC-excHC Barring of outgoing international calls except those directed to the home PLMN country, subgroup of call barring SS BARG Billing and Accounting Rapporteur Group, a working group within the GSM MoU looking after billing and accounting issues BB
Baseband
BCCH Broadcast control channel, channel on which the network supplies the mobile station with information on parameters currently used in the network BCF
Base control function
BDN Barred dialing numbers, elementary file on the SIM card that is organized similar to ADN and contains numbers that are barred BER
Bit error ratio
BiCMOS
Bipolar complementary metal-oxide semiconductor
BICroam Barring of incoming calls when roaming outside the home PLMN, subgroup of call barring SS BOAC BOC BOIC SS BS
Barring of outgoing calls, subgroup of call barring SS Bell Operating Company Barring of outgoing international calls, subgroup of call barring
Base station
BSC
Base station controller, control entity for the base station
BSS
Base station subsystem, made up of BSC and BTS
480
BT
GSM and Personal Communications Handbook
Bandwidth (B) multiplied by bit period (T)
BTA
Basic Trading Area
BTS
Base transceiver station
BZT Bundesamt für Zulassung in der Telekommunikation, German approval body for telecommunications C/I
Carrier-to-interference ratio
CAI (1) Charge advice information, set of parameters used by the mobile station to calculate the charging information; (2) common air interface CAMEL Customized applications for mobile network enhanced logic, intelligent network and tool for network operators to make their features available to subscribers roaming abroad CB
Call barring
CB Cell broadcast, allows transmission of short information from the network to the mobile station within a geographical area CBC Cell broadcast center, schedules the transmission of cell broadcast messages within a network CBCH service
Cell broadcast channel, physical channel used by cell broadcast
CB-DRX Cell broadcast discontinuous reception, new Phase 2+ feature allowing the scheduling of the reception of new information coming from the network CBE
Cell broadcast entity, originates cell broadcast messages
CBMI Cell broadcast message identifier selection, elementary file on SIM used in conjunction with cell broadcast CBMID Cell broadcast message identifier for data download, elementary file on SIM used in conjunction with SIM application toolkit CBMIR Cell broadcast message identifier range selection, elementary file on SIM used along with cell broadcast
Glossary
481
CBS Cell broadcast short message service, allows the transmission of short information units from the network to the mobile station within a geographical area CC
Convolutional code
CCBS Completion of call to busy subscriber, call completion supplementary service CCCH
Common control channel
CCIR
International Radio Consultive Committee
CCITT International Committee CCM
Telegraph
and
Telephone
Consultative
Current call meter, elementary file on the SIM
CCNRc Completion to call when subscriber not reachable, call completion supplementary service CCNRy Completion to call when no reply, call completion supplementary service CCP
Capability configuration parameters, elementary file on the SIM
CCS
Call century seconds
CDCS
Continuous dynamic channel selection
CDG
CDMA Development Group
CDL
Coded DCCH locator (IS-136)
CDMA CDVCC CELP
Code division multiple access Control digital verification color code (IS-136) Code(-book) excited linear predictive (speech codec)
CEPT European Conference of Posts and Telecommunications Administrations (Conférence Européenne des Administration des Postes et des Télécommunication) CF
Call forwarding, call offering supplementary service
482
GSM and Personal Communications Handbook
CFB Call forwarding on mobile busy subscriber, call offering supplementary service CFNRc service
Call forwarding on not reachable, call offering supplementary
CFNRy service
Call forwarding on no reply, call offering supplementary
CFU Call forwarding unconditional, call offering supplementary service CH
Call hold
CHV1 Card holder verification 1, parameter on SIM usually referred to as the PIN1 CHV2 Card holder verification 2, parameter on SIM usually referred to as PIN2 CIS
Commonwealth of Independent States
CLIP Calling line identification presentation, line identification supplementary service CLIR Calling line identification restriction, line identification supplementary service CMOS CNL CODEC
Complementary metal-oxide semiconductor Cooperative network list, elementary file on the SIM Coder-decoder
COLP Connected line identification presentation, line identification supplementary service COLR Connected line identification restriction, line identification supplementary service CPU
Central processing unit
CRC
Cyclic redundancy code/check
CS Coding scheme, for example, used for different channel coding for GPRS to achieve different (higher) data rates
Glossary
483
CSAIG Central/Southern Africa Interest Group, an interest group within the GSM MoU CSE
CAMEL service environment, logical entity supporting CAMEL
CSFP
Coded superframe phase (IS-136)
CSG Communication strategy group, a working group within the GSM MoU CSI CAMEL subscription information, subscriber-related information stored in the network CSIC
Customer-specific integrated circuit
CSPDN
Circuit-switched public data network
CT (1) Call transfer, call completion supplementary service; (2) cordless telephone CTIA
Cellular Telecommunications Industry Association (USA)
CTM
Cordless telephone mobility
CUG
Closed user group, supplementary service
CW D/A
Call waiting, call completion supplementary service Digital-to-analog conversion
DA Destination address, within a short message, usually a phone number of the addressee DAC
Digital-to-analog converter
dBm
Decibels relative to 1 mW
DCA
Dynamic channel allocation
DCCH DCK
Digital control channel (IS-136) Depersonalization control keys, elementary file on SIM
DCS Data coding scheme, of the short message content; (2) digital cellular system DD
Differential decoder (PACS)
484
GSM and Personal Communications Handbook
DECT DF
Digital Enhanced Cordless Telecommunications Dedicated file, directory on the SIM
DFE
Decision feedback equalizer
DGPT
Directorate General of Post and Telecommunication (France)
DISC
Disconnect, command for the radio link protocol
DLC DM
Data link control (Layer) Disconnect mode, a state of the radio link protocol
DQPSK
Differential quadrature phase shift keying
DRX
Discontinuous receive
DSA
Direct subscriber access, supplementary service
DSAR
Direct subscriber access restriction, supplementary service
DS-CDMA DSP
Direct sequence code division multiple access
Digital signal processor
DSSS
Direct sequence spread spectrum
DTC
Digital traffic channel (IS-136)
DTE Data terminal equipment, equipment that is normally connected to a MS for the exchange of data DTI
Department of Trade and Industry, United Kingdom
DTMF DTX EBCCH
Dual-tone multifrequency Discontinuous transmission Extended broadcast control channel (IS-136)
EBRC EMC Bio-Effects Review Committee Group, a working group within GSM MoU ECAIG East Central Asia Interest Group, an interest group within the GSM MoU ECC
Emergency call codes, elementary file on SIM
Glossary
485
ECT
Explicit call transfer, supplementary service
EEC
European Economic Community
EEPROM EF
Electrical erasable and programmable read-only memory
Elementary file, data record on SIM
EFR GSM
Enhanced full-rate speech codec, improved speech codec used in
EIA
Electronics Industry Association (USA)
EIG
European Interest Group, an interest group within the GSM MoU
EMC
Electromagnetic compatibility
eMLPP Enhanced multilevel precedence and preemption service, supplementary service EMR EP
Electromagnetic radiation Error profile
ESN
Electronic serial number
ETS
European Telecommunications Standard
ETSI
European Telecommunications Standards Institute
EVRC
Enhanced variable rate (speech) codec
FBCCH FC
Fast broadcast control channel (IS-136)
Fast channel (PACS)
FCC FCCH
Federal Communications Commission (USA) Frequency correction channel
FCS Frame check sequence, portion of a frame used by the radio link protocol to detect errors FDD FDMA
Frequency division duplex Frequency division multiple access
486
GSM and Personal Communications Handbook
FDN Fixed dialing numbers, elementary file on the SIM card that is organized similar to ADN and contains numbers to which the phone’s use can be restricted FEC FF
Forward error correction Fraud Forum, a working group within the GSM MoU
FM
Frequency modulation
FOCC FP
Forward control channel (IS-136/AMPS) Fixed part (base station)
FPLMN
Forbidden PLMN, filed on the SIM
FPLMTS Future Public Land Mobile Telecommunications System, a proposal for third-generation cellular system, see IMT-2000 FR
Full-rate speech codec, standard speech codec used in GSM
FTA
Full type approval
FVC
Forward voice channel (IS-136/AMPS)
GaAs GAP
Gallium arsenide Generic access profile
GCR Group call register, register in the network which is used with voice group call service GFSK
Gaussian-filtered frequency shift keying
GGSN
Gateway GPRS support node, gateway entity for GPRS support
GID1 Group identifier level 1, elementary file on the SIM used for personalisation GID2 Group identifier level 2, elementary file on the SIM used for personalization GMSC
Gateway mobile switching center
GMSK
Gaussian-filtered minimum shift keying
GPRS GSM
General packet radio service, packet data service defined for
Glossary
487
GPS Global positioning system, a satellite system used for accurate positioning information GSM
Global system for mobile communications
gsmSCF GSM service control function, function in the network related to CAMEL gsmSSF GSM service switching function, function in the network related to CAMEL HLR Home location register, network register for home subscribers, always associated with a MSC HOLD
Call hold, supplementary service
HPLMN
HPLMN search period, elementary file on SIM
HR Half-rate speech codec, optional speech codec used in GSM, allowing for increased system capacity HSCSD High-speed circuit-switched data, data service defined for GSM allowing the combining of up to 8 time slots for transmission or reception IC
Integrated circuit
ICCID ICM IF
ICC identification, elementary file on SIM In-call modification
Intermediate frequency
IIG
Indian Interest Group, an interest group within the GSM MoU
IM
Intermodulation
IMSI International mobile subscriber identity, elementary file on SIM holding the unique number identifying the subscriber within a GSM network IMT-2000 International Mobile Telecommunications 2000 System, replacement for the FPLMTS designation, see FPLMTS IN Intelligent networks, CAMEL being an example for IN as used in GSM
488
GSM and Personal Communications Handbook
I/O
Input/output
IP
Internet protocol
IP3
Intercept point of the third order
IP-M IP multicast, transmission of packet data (GPRS) from one subscriber to multiple subscribers via the Internet protocol IPR
Intellectual property right
IREG International Roaming Expert Group, a working group within the GSM MoU looking after international roaming issues IS
Interim standard (US)
ISDN
Integrated services digital network
ISI
Intersymbol interference
ITA
Interim type approval
ITU
International Telecommunication Union
IWF Interworking function, interface between cellular network and outside PSTN or data network JTC
Joint Technical Committee (TIA body)
Kbps
Kilobits per second (1000 bits per second)
KB
Kilobyte
Kc
Ciphering key
Ki
Internal key
Ksps L&RG MoU LAI LAN
Kilosamples per second (1,000 samples per second) Legal and Regulatory Group, a working group within the GSM Location area information Local-area network
LAPM Link access procedure for modems, protocol for data exchange between modems
Glossary
489
LEC
Local exchange center, local switch in a PSTN or ISDN
LEO
Low-earth orbit
LFSR
Linear feedback shift register
LNA
Low-noise amplifier
LND
Last numbers dialed, elementary file on the SIM
LO
Local oscillator
LOCI
Location information, elementary file on the SIM
LP
Language preference, elementary file on the SIM
LP
Linear prediction/predictive (speech coding)
LPC
Linear predictive (speech) coding
mA
Milliampere
MAC
Medium access control (layer)
MAH
Mobile access hunting, supplementary service
mAh
Milliampere-hour
MCC
Mobile country code
MCID ME MEO MF mips
Malicious call identification, supplementary service Mobile equipment (GSM terminal without SIM) Medium-earth orbit Master file, main directory on SIM Millions of instructions per second
MLPP Multilevel precedence and preemption service, supplementary service in ISDN MLSE MMI MMIC
Maximum likelihood sequence estimation Man/machine interface, also called user interface Monolithic microwave integrated circuits
490
GSM and Personal Communications Handbook
MMS More-messages-to-send, status flag within the short message to indicate that more messages are scheduled to be sent MNC
Mobile network code
MNP Microcom networking protocol, data protocol to secure data transmission and compression of data between two modems MO MOS
Mobile originated (1) Metal-oxide semiconductor; (2) mean opinion score
MoU Memorandum of Understanding, within a group of GSM operators MPT MPTY
Ministry of Posts and Telecommunications (Japan) Multiparty communication, supplementary services
MR Message reference, mobile station internal referencing of messages to which the network also makes reference ms
Millisecond, 1/1,000 of a second
MS
Mobile station (in GSM, terminal with SIM)
MSC Mobile services switching center, switch within the GSM network MSIN Mobile subscriber identification number, portion of the IMSI within a dedicated network MSISDN Mobile station international ISDN number, international phone number of a GSM subscriber MSK
Minimum shift keying
MSP
Multiple subscriber profile, supplementary service
MT MTA
Mobile terminal/termination Major Trading Area
MTI Message type indicator, indicating for short messages the purpose of the sent message MTSO
Mobile telephone switching office
Glossary MUX
491
Multiplexer
NAIG North American Interest Group, an interest group within the GSM MoU NB
Narrowband
NCH service
Notification channel, logical channel used for voice group call
NDUB
Network-determined user busy
NITZ Network identity and time zone, service defined for GSM networks to transmit identity and time zone regularly through the network NMT ns
Nordic mobile telephone Nanosecond, 1/1,000,000,000 of a second
NT Nontransparent services, data service that uses the radio link protocol between the MS and the MSC NTA
National Telecom Agency (Denmark)
NTP
Nominal transmitted power
NTT
Nippon Telephone and Telegraph Company
NULL
Null information, frame used for the radio link protocol
OA Originating address, details, for example, the phone number of the sender ODB
Operator-determined barring
OQPSK
Offset quadrature phase shift keying
OSI
Open systems interconnection
OSS
Operator-specific services
OVHM
Overhead message (IS-136)
P25
Project 25 (formerly APCO Project 25)
PA
Power amplifier
PACS
Personal Access Communications System
492
GSM and Personal Communications Handbook
PAD Packet assembler/disassembler, access to a packet data network coming from a circuit-switched network, such as GSM PAMR
Public access mobile radio
PBX
Private branch exchange
PCC
Power control channel (PACS)
PCH
Paging channel (IS-136)
PCM
Pulse code modulation
PCN
Personal communications network, also referred to as DCS 1800
PCNIG Personal Communications Networks Interest Group, a technology group within the GSM MoU PCP
Power control pulse (IS-661)
PCS Personal communications system, also referred to as PCS 1900 or GSM NA PDA
Personal digital assistant (palmtop organizer/computer)
PDC
Personal digital cellular
PDN
Packet data network
PDS Packet data on signaling channels, implementation of packet data services for GSM PDSS2-SN PDS PFC
PDSS2-support node, support node for second service of
Paging frame class (IS-136)
PH Packet handler, access to a packet data network coming from a circuit-switched network such as GSM PHP
Personal handy phone
PHS
Personal Handy Phone System
PID Protocol identifier, identifies to which protocol a short message should be transferred PIN
Personal identification number
Glossary PLL
493
Phase-locked loop
PLMN
Public land mobile network, cellular network
PLMNsel PMR
(1) Private mobile radio; (2) professional mobile radio
POTS PP
PLMN selector, elementary file on SIM
Plain old telephony service Portable part (mobile station)
ppm PRBS pr-ETS PRS PSID PSPDN
Parts per million Pseudorandom bit sequence Preliminary European telecommunications standard Premium rate services Private system identifier (IS-136) Packet-switched public data network
PSTN Public-switched network
telephone
network,
standard
wireline
PTM Point-to-multipoint, transmission between one single point subscriber or base station to several subscribers; this applies for packet data services or short message service PTM-G PTM group call, point-to-multipoint service for general packet radio service allowing the transmission of data to a predefined group PTM-M PTM multicast, point-to-multipoint service for general packet radio service allowing the transmission of data to multiple subscribers within an area PTP Point-to-point, transmission between two dedicated subscribers; this can apply to packet data services or short message service PTT PUCT PUK
Push-to-talk Price per unit and currency table, elementary file on SIM Personal unblocking key (for GSM SIM)
494
GSM and Personal Communications Handbook
PWT
Private wireless telecommunications
QCELP
Qualcomm codebook excited linear predictive
QCPM
Quadrature continuous-phase modulation
QPSK
Quadrature phase shift keying
RA Rate adaptation, data transmission speed adaptation between different interfaces RACE Research and Development of Advanced Communications Technologies and Services (in Europe) RACH RAM
Random access channel Random access memory
RAND Random value, used for authentication and calculation of the ciphering algorithm RDI Restricted digital information, direct digital data exchange between a GSM and a digital network RDS Radio data system, enables the display of characters on the radio’s display RECC REJ RF RISC RLL
Reverse control channel (IS-136/AMPS) Reject, layer 2 frame/information Radio-frequency Reduced instruction set computer Radio local loop
RLP Radio link protocol, link protocol to reduce the error rate via the air interface RLR
Receive loudness ratio
RNR
Receive not ready, layer 2 frame/information
ROM
Read-only memory
Glossary
495
RP (1) Reply path, feature within SMS that allows the sending of replies to messages free of charge for the sender; charge will be picked up by the recipient; (2) radio port (PACS) RPCU
Radio port controller unit (PACS)
RPE-LTP RR
Regular pulse-excited–long-term prediction
Receive ready, layer 2 frame/information
RSID
Residential system identifier (IS-136)
RTOS
Real-time operating system
RVC RX
Reverse voice channel (IS-136/AMPS) Receiver
SABM Set asynchronous balanced mode, a state of the layer 2 or the radio link protocol SACCH Slow associated control channel, a logical channel type containing regular control messages between the network and the mobile station SAPI Service access point identifier, identifier for different logical access points SAW
Surface acoustic wave (ceramic-based filter)
SC (1) Service center for short message services; (2) service code, used for supplementary services; (3) slow channel (PACS) SCAG
Smart card application group, a group within the GSM MoU
SCF
Shared channel feedback (IS-136)
SCH
Synchronization channel
SCT
Single-step call transfer
SCTP Service center time stamp, time stamp for a short message that arrives at the service center SDCCH SDN
Stand-alone dedicated control channel, signaling channel Surface dial numbers
496
GSM and Personal Communications Handbook
SEQAM (IS-661) SERG
Spectrally efficient quadrature amplitude modulation Services Expert Rapporteur Group, GSM MOU subgroup
SG Security Group, a working group within the GSM MoU looking after security issues SGSN Serving GPRS support node, entity needed to support general packet radio service SI
Supplementary information, used for supplementary services
SIM
Subscriber identity module
SLR
Sending loudness ratio
SME Short message entity, entity originating a short message, for example, a mobile station SMG SM-MT SMR
Special mobile group Short message–mobile terminated Specialized mobile radio
SMS Short message service, service that allows the transmission of messages up to 160 characters SMSCH
SMS message channel (IS-136)
SMS-GMSC Short message service–gateway mobile services switching center, logical entity connected to the MSC that is used to coordinate the delivery of SMS SMS-IWMSC Short message service–interworking mobil services switching center, local entity connected to the MSC which is used to coordinate originating SMS SMSP
Short message service parameter, elementary file on SIM
SMSS
SMS status, feature of SMS
SNR
Signal-to-noise ratio
SOR
Support of optimal routing
Glossary
497
SPACH SPN
SMS/PCH/ARCH channel (IS-136) Service provider name, elementary file on SIM
SPNP
Support of private numbering plan, supplementary service
sqkm
Square kilometer, area of 1 km × 1 km
SREJ
Selective reject, layer 2 frame/information
SRES
Authentication result
SRI Status report indication, indication that a status report will be issued for this message SRR Status report request, indication that a status report is requested for this message SS
Supplementary services
SSB
Single sideband
SS No. 7
Signaling System Number 7
SST SIM service table, elementary file on SIM describing the services supported by a SIM SU
Subscriber unit (PACS)
T Transparent, data service that does not use the radio link protocol between the MS and the MSC T1P1 TA
Network Interfaces Committee of ANSI (USA) Type approval; terminal adapter
TADIG Transferred Account Data Interchange Group, a working group within GSM MoU TAF Terminal adaptation function, dedicated function between the mobile station and the data terminal for data/fax transmission TBR
Technical basis for regulation
TDD
Time division duplex
TDMA
Time division multiple access
498
TE
GSM and Personal Communications Handbook
Terminal equipment
TETRA radio)
Terrestrial trunked radio (formerly trans-European trunked
TFO
Tandem-free operation
TIA
Telecommunications Industry Association (USA)
TMSI TRAU TWG MoU
Temporary mobile subscriber identity Transcoder rate adaptation unit Terminal Working Group, a working group within the GSM
TX
Transmitter
UA
Unnumbered acknowledge, layer 2 frame/information
UCI
Universal computer interface
UCS2 Universal Coding Scheme 2, 16-bit based coding scheme defined for SMS to support, for example, Arabic and Chinese UD
User data, actual message that is to be transmitted with the SMS
UDI Unrestricted digital information, direct digital data exchange between a GSM and ISDN network UDL User data length, the length of the user data to be transmitted with the SMS UDUB
User-determined user busy
UI (1) Unnumbered information, layer 2 frame/information; (2) user interface UIC Union Internationale de Chemin de Fer, an organization of European Railways UMTS Universal Mobile Telecommunication System, ETSI standardization proposal for third-generation cellular system UPT
Universal personal telecommunications
Glossary USSD service
499
Unstructured supplementary services data, supplementary
UUI User-to-user information, data transmitted by user to user signaling UUS
User-to-user signaling, supplementary service
UWCC IS-41) VA
Universal Wireless Communications Consortium (IS-136/ Viterbi algorithm
VAD
Voice activity detection
VBS Voice broadcast service, supplementary service and elementary file on SIM VBSS
Voice broadcast service status, elementary file on SIM
VCO
Voltage-controlled oscillator
VGA
Variable gain amplifier
VGCS Voice group call service, supplementary service and elementary file on SIM VGCSS VGS
Voice group call service status, elementary file on SIM Voice group service
VLR Visitor location register, network register for visiting subscribers, always associated with a MSC VP
Validity period of a sent short message
VPF Validity period format, the format of the validity period, either absolute or relative VPN
Virtual private network
VSELP
Vector sum excited linear predictive (IS-136)
WACS
Wireless Access Communications System
WAFU
Wireless access fixed unit (PACS)
WB
Wideband
500
GSM and Personal Communications Handbook
WCDMA WIN WLAN
Wideband CDMA
Wireless intelligent network (usually refers to IS-41/IS-136) Wireless local-area network
WLL
Wireless in the local loop
WTR
Wireless Technology Research
XID
Exchange identification, layer 2 frame used by the RLP
About the authors
About the authors Siegmund M. Redl graduated from the Technical University of Munich in 1989 with a degree in communications engineering. He joined Schlumberger Technologies, Communications Test Division (known as Wavetek since 1994), Munich, in early 1990 as a support engineer for GSM Test Systems. He continued to hold positions within Schlumberger in marketing and product management for digital wireless communication test systems as he participated in the ETSI GSM standardization efforts. In 1995 he joined LSI LOGIC, a leading supplier of integrated circuits, where he is director of the Wireless Communications Business Unit. Matthias K. Weber graduated from the Technical University of Munich in 1989 with a degree in communications engineering. He joined Schlumberger Technologies, Communications Test Division (known as Wavetek since 1994), Munich, in October 1989. After a period in R&D on Schlumberger’s GSM test equipment, he turned to providing technical product support for GSM radio test systems before he assumed responsibility for product management of GSM/DCS-1800 test equipment. In 1995 he joined Sony Personal Communication Europe as a technical product manager. He now is a manager in the Product Planning group for GSM-related products for Sony PCE. Malcolm W. Oliphant graduated from Hawaii Loa College, Kaneohe, Hawaii, with a degree in biology. During the next 20 years he worked in a 501
502
GSM and Personal Communications Handbook
variety of technical fields, including avionics, metrology, and mobile radio systems engineering. He joined Schlumberger Technologies, Communications Test Division (known as Wavetek since 1994) in early 1990 where he worked in marketing and applications engineering. In late 1996 he joined IFR Systems, Wichita, Kansas, where he manages the marketing, applications engineering, and training departments. He is active in North American GSM and public safety radio standards committees.
Index # (pound sign), 317 * (asterisk), 317 + (plus character), 317 3rd Generation Interest Group (3GIG), 39, 65
A A5 encryption algorithm, 461 Abbreviated dialing numbers (ADN), 317–18 Abis network interface clock, 467–68 Accelerators, 414 Access control class EF, 315 Access grant channel (AGCH), 396 Access response channel (ARCH), 108 Accumulated call meter (ACM), 277, 313–14 Adaptive differential pulse code modulation (ADPCM), 29, 35, 87, 417 Advanced multirate (AMR), 426, 428 Advice of charge (AoC), 136, 375–79
accumulated call meter (ACM), 277 AoCC, 276, 278–79 AoCI, 276, 277–78 charge advice information (CAI), 276–77 current call meter (CCM), 277 defined, 275 obstacles, 298 price per unit and currency table (PUCT), 277 See also Supplementary services (SS) Algebraic codebook excited linear predictive (ACELP), 106, 426 American National Standards Institute (ANSI), 61 Amplification IF, 408 RF, 407–8 transmitter, 401 AMPS, 15, 103 digital (D-AMPS), 7 interworking, 15 markets, 103
503
504
GSM and Personal Communications Handbook
AMPS (continued) plus-CDMA, 24 plus-TDMA, 24 Analog modems, 173 Analog-to-digital converter (ADC), 392, 409 Analysis by synthesis (AbS), 419 APCO 25 system, 119 Asia wireless technology, 12–13 Asynchronous data rates, 159 AT command set, 191–93 call control commands, 192 general commands, 192 mobile equipment control and status commands, 193 mobile equipment errors, 193 network service related commands, 192–93 Australia AMPS system, 8 Authentication algorithms, 459–60 Authentication keys, 459 Authentication result (SRES), 306 generating, 307 procedure for passing, 308 Automatic frequency control (AFC), 464 Automatic gain control (AGC), 464 Automatic repeat requests (ARQs), 153
B Barred dialing numbers (BDN), 140 EF, 328 restrictions, 332 Barring of all incoming calls (BAIC), 263 Barring of incoming calls (BIC), 270 Barring of incoming calls when roaming outside the home PLMN (BICroam), 260, 263
Barring of outgoing calls (BOAC), 260, 263 Barring of outgoing international calls (BOIC), 260, 263 Baseband codec, 395 implementation, 412 signal processing, 412–14 Base station controllers (BSCs), 112, 196–97, 239–40, 373 defined, 239 responsibilities, 240 supplementary services and, 251 Base stations (BSs) number of, 411 physical/logical blocks of, 396–97 signal path, 397 subsystem (BSS), 184, 215 timing reference, 467 Base transceiver station (BTS), 251, 372 ciphering algorithm implementation in, 460 transmitters, 398 Basic service groups defined, 245 list of, 248–49 See also Bearer services; Teleservices Basic Trading Area (BTA), 45, 62 Bearer services, 130, 149–51 codes for, 286 defined, 148 general, 205–6 illustrated, 148 list of, 150 nontransparent, 132, 149 Phase 1, 132, 134 Phase 2+, 205–6 Release 96, 139, 142 Release 97, 143 transparent, 132, 149
Index
See also Teleservices Billing and Accounting Rapporteur Group (BARG), 66 Billing center, 373 Bit error rates (BERs), 35 Broadcast control channel (BCCH),108, 238, 396 EF, 314 messages, 108
C Call barring (CB), 133, 262–65 applicability of, 264 BAIC, 263 BIC-roam, 260, 263 BOAC, 260, 263 BOIC, 260, 263 BOIC-excHC, 263 defined, 262 incoming, restrictions, 265 for incoming and outgoing calls, 263–64 outgoing, restrictions, 265 password change, 288 See also Supplementary services (SS) Call completion services, 292–94 completion of call to busy subscriber (CCBS), 293–94 example of, 294 principle operator for, 294 Call deflection, 291 Call forwarding (CF), 133, 247, 252–62 application of, 253 behavior of, 253–55 conditional, 257–59 conflicts, 259–60 deactivation of, 254 defined, 252
505
endless, 255 enhancements, 291 erasure procedure, 254 forwarding-to number, 254 to home country, 382–83 interrogation procedure, 254 on mobile busy subscriber (CFB), 253, 258 on no reply (CFNRy), 253, 259 on not reachable (CFNRc), 253, 259 operation of, 255–56 payment for, 260–62 registration, 253, 254, 287 unconditional (CFU), 253, 256–57, 260 to visited country, 384 See also Supplementary services (SS) Call holding (HOLD), 271–72 aspects of, 272 availability of, 271 defined, 271 options, 271 See also Supplementary services (SS) Calling line identification, 135, 266–68 activation of, 286–87 call flow, 267 CLIP, 266, 268 CLIR, 267 See also Supplementary services (SS) Calling line identity (CLI), 72 Call-related supplementary services control codes and functions of, 289 defined, 284 implementation of, 288 See also Supplementary services (SS)
506
GSM and Personal Communications Handbook
Call screening, 288 Call transfer (CT), 291–92 changing scenarios, 292 defined, 291 explicit (ECT), 292 single-step (SCT), 292 See also Supplementary services (SS) Call waiting (CW), 136, 268–71 activation of, 269, 286 BIC conflict with, 270 deactivation of, 269 defined, 268 operation of, 269–70 procedures for, 270 See also Supplementary services (SS) CDMA Development Group (CDG), 102 Cell broadcast messages, 238 contents of, 240–43 data coding scheme, 242 display for zone indication, 243 format, 241 geographical scope, 241, 242 identifier for data download, 327 identifier range selection, 327 message code, 241 message identifier, 242, 351 page parameter, 242 serial number, 241 update number, 241 Cell broadcast SMS, 237–43 capacity, 243 center (CBC), 239, 240 channel (CBCH), 238 data download via, 351 defined, 211, 237–38 discontinuous reception (CB-DRX), 243 entity (CBE), 239 future developments for, 243
implementation in network, 238–40 network architecture for, 239 RDS vs., 237–38 service (CBS), 138, 238, 240 See also Short message services (SMS) Cellular, 4–5 analog vs. digital subscribers, 7 CDMA, 15, 16 custom base profile, 18 infrastructure costs, 16 market presence and potential, 10–13 meeting demands of, 13–15 PCS vs., 4 penetration, 6 subscribers worldwide, 11 United States, 43 Cellular phones. See Mobile terminals Cellular Telecommunications Industry Association (CTIA), 78–79 Central processing unit (CPU), 336 Channel coding, 394, 439 decoding, 394 estimate, 446 Charge advice information (CAI), 276–77 Charging for call forwarding, 380 exceptions, 380–81 for HSCSD, 199 for MPTY, 275 for national calls, 378–79 principles, 377–81 when roaming, 379–80 for supplementary services, 298–99
Index
for unconditional call forwarding, 261 Checkpointing, 171 Chip cards, 303 Ciphering, 460–61 algorithms, 460 sequence, 460, 461 Ciphering key (Kc), 306 EF, 311 generating, 307 procedure for passing, 308 Circuit-switched public data network (CSPDN), 151 interconnection methods, 177–78 transmission into, 177–78 Closed user group (CUG), 120, 136, 279–81 availability, 280 barring services, 281 defined, 279 index, 280 subscription options, 280–81 See also Supplementary services (SS) Coded DCCH locator (CDL), 106 Code division multiple access (CDMA), 7, 15, 101–4 advantages, 9 cellular, 15, 16 channel loading, 33 growth, 10 IS-95, 17 maturity, 102–3 operators, 48 preferred frequency assignments, 102 technology, 9 wideband, 42 See also IS-95 Coded superframe phase (CSFP), 107 Code excitation linear prediction (CELP), 419, 421
507
algebraic (ACELP), 106, 426 codecs, 421 Coders hybrid, 419–22 vocoders, 417–19 waveform, 416–17 See also Speech coding Coding schemes (CS), 202 Combining and splitting functions, 197, 199 Common air interface (CAI) standard, 42 Common control channels (CCCH), 396 Companding, 393 Complementary metal-oxide semiconductor (CMOS) processes, 463 Completion of call to busy subscriber (CCBS), 143, 293–94 Complex numbers, 434 Conditional call forwarding, 257–59 conditions, 258–59 defined, 257 operation of, 258 payment for, 262 See also Call forwarding (CF) Connected line identification, 135, 268 COLP, 268 COLR, 268 information flow, 269 See also Supplementary services (SS) Continuous channel selection (CDCS), 93 Control digital verification color code (CDVCC), 107 Convolutional decoders, 448–58 Convolutional encoders, 446–48 codes, 448 defined, 446–47
508
GSM and Personal Communications Handbook
Convolutional encoders (continued) illustrated, 447 Cooperative network list, 327 Cordless telephone (CT), 30, 84–85 Cordless telephony mobility (CTM), 89 Coverage, network, 29, 32–34 Current call meter (CCM), 277 Customized applications for mobile network enhanced logic (CAMEL), 141, 357–61 defined, 345, 358 example, 360–61 functional description of, 358–59 implementation, 359 network architecture, 360 service environment (CSE), 358 subscribers, 359 subscription information (CSI), 358 Cyclic redundancy check (CRC), 100, 394
D Data rates 14.4-Kbps user, 204 asynchronous, 159 nontransparent fax, 185 synchronous, 159, 161 Data terminal equipment (DTE), 187 DCS 1800 defined, 56 history of, 57 operators, 58 specifications, 56 See also Personal communications network (PCN) Decision feedback equalizers (DFE), 443 Decoding
channel, 394 incorrect, in presence of too many errors, 454 in presence of errors, 452 rules and example, 449 soft decision, 454, 455, 457 Demodulators, 408 Depersonalization control keys (DCK), 327 Deregulation, 14–15, 18, 70 Developing countries, 13–15 Differential decoder (DD), 100 Digital AMPS (D-AMPS), 7 Digital control channel (DCCH), 105 bursts, 107 coded locator (CDL) channel, 106 time slots, 108 Digital enhanced cordless telephone (DECT), 23, 88–96 applications, 96 channel spacing, 90 DCA, 33, 93 defined, 88 dual-mode operation, 95 FM deviation, 92 GSM combination, 25 interworking, 95 open standard, 89 protocol stack, 95 RF pulse shape, 93 signaling, 94–95 standard, 88 TDD/TDMA frame/slot structure, 91 technical comparison, 87–88 terminals, 89 Digital signal processors (DSPs), 413, 414, 468 Digital-to-analog converter (DAC), 392 Digital traffic channel (DTC), 105
Index
Direct sequence spread spectrum (DSSS), 116 Direct subscriber access (DSA), 295 Direct subscriber access restriction (DSAR), 295 Discontinuous receive (DRX), 410 Discontinuous transmission (DTX), 17, 171, 394 Distinctive ringing, 290 Distribution, 19 Diversity frequency, 439 space, 439 time, 440 Dual-tone multifrequency (DTMF) decoders, 424 defined, 423 handling in GSM, 423–24 messages, 423 signaling tones and, 424 Duplexers, 401, 407 Dynamic channel allocation (DCA), 33, 87, 93
E Eastern Europe wireless technology, 13 Echo cancellation, 431–33 Echo suppression, 430–31 Electrical erasable and programmable read-only memory (EEPROM), 304, 334, 337 Electromagnetic compatibility (EMC), 122 Electromagnetic radiation (EMR), 122 Electronic serial numbers (ESNs), 462 Elementary files (EF), 305, 310–23
509
abbreviated dialing numbers, 317–18 access control class, 315 accumulated call meter, 313 ACM max value, 313–14 addition of, 343 administrative data, 316 automatic answer of eMLPP, 326 barred dialog numbers, 328 broadcast control channel, 314 capability configuration parameters, 321 cell broadcast message identifier for data download, 327 cell broadcast message identifier range selection, 327 cell broadcast message identifier selection, 314 ciphering key, 311 cooperative network list, 327 depersonalization control keys, 327 emergency call codes, 326 enhanced multilevel precedence and preemption, 326 extension 1, 323 extension 2, 323 extension 3, 328 extension 4, 328 fixed dialing numbers, 318–19 forbidden PLMN, 315 group identifier level 1 and 2, 324–25 HPLMN search period, 312 ICC identification, 310 IMSI, 311 language preference, 311 last number dialed, 322 list of, 328–32 location information, 315–16 mobile station international ISDN number, 321–22
510
GSM and Personal Communications Handbook
Elementary files (EF) (continued) phase identification, 316 PLMN selector, 311–12 price per unit and currency table, 314 service dialing numbers, 327–28 service provider name, 325 short messages, 319–20 short message service parameter, 320–21 SIM service table, 313, 323–24 SMS status, 321 voice broadcast service, 325–26 voice broadcast service status, 326 voice group call service, 325 voice group call service status, 325 See also Subscriber identity module (SIM) Emergency call service, 206 Encryption, 459–62 Endless call forwarding, 255 Enhanced full-rate (EFR), 61, 74, 425–27 algorithm, 394 complexity comparison, 427 PCS 1900 and, 135 specifications, 425–26 vocoders, 35 voice quality, 426 See also Speech coding Enhanced multilevel precedence and preemption (eMLPP) service, 64, 139, 361–65 automatic answer for, 326, 363 call setup time improvements, 362–63, 364 examples, 365 precedence service, 361 preemption service, 361 priority levels, 361–62 subscribers, 362
Enhanced variable rate codec (EVRC), 426 Equalizers, 433–59 decision feedback (DFE), 443 defined, 434, 441 feedback, 443 general, 441–44 ISI and, 433, 435–41 linear, 442 parts of, 441 power and, 434 Viterbi, 434, 444–59 Erlangs, 33–34, 94 Error profiles (EPs), 423 Europe ACTS program, 38 cellular, 12, 13 RACE programs I and II, 38 UMTS, 38–40 European Economic Community (EEC), 78 European Telecommunications Standard (ETS), 426 European Telecommunication Standards Institute (ETSI), 38, 39, 67–69 history of, 67 See also Special Mobile Group (SMG) Exchange identification (XID), 169–70 defined, 169 parameters, 169–70 Explicit call transfer (ECT), 139, 292
F Facsimile transmission, 174–75 Fax adapter, 181 Fax services, 178–86 end-to-end view, 179–80
Index
enhancements, 204–5 in-call modification (ICM), 186 mobile station configuration, 181–82 network infrastructure support, 180 nontransparent, 179, 185–86 transparent, 179, 182–85 Federal Communications Commission (FCC), 41, 46 Feedback equalizers, 443 Field testing, 76 Filters backward, 443 image, 408 matched, 442 narrow, 439 preselector, 407 SAW, 403 transmitter, 401 Fixed dialing numbers (FDN), 318–19 ADN vs., 318–19 deactivating, 319 operation for, 320 restrictions, 332 Fixed parts (FPs), 90 Forbidden PLMN, 315 Forward control channel/reverse control channel (FOCC/RECC), 105 Forward error correction (FEC), 153, 161, 163 as envelope, 164 HSCSD, 198 Frame check sequence (FCS), 167 Frequency correction channel (FCCH), 396 Frequency diversity, 439 Frequency hopping, 195 Frequency synthesizers, 400 Fulfillment houses, 20
511
Full-rate (FR) speech coding, 74, 393, 424 complexity comparison, 427 specifications, 424 See also Speech coding Full-slot burst, 91 Full type approval (FTA), 76 Future Public Land Mobile Telecommunications System (FPLMTS), 38
G Gateway GPRS support node (GGSN), 200, 201 Gateway mobile services switching center (GMSC), 249, 374 Gaussian-filtered frequency shift keying (GFSK), 91, 92 General packet radio service (GPRS), 143, 199–202 coding schemes (CS), 202 GGSN, 200, 201 implementation on radio interface, 201–2 introduction of, 202 network architecture for, 200–201 point-to-multipoint (PTM) service, 200 point-to-point (PTP) service, 200 SGSN, 200–201 Generic access profile (GAP), 89 Geographical scope, 241 Global positioning system (GPS), 25, 206 Global roaming, 25 Global System for Mobile Communications. See GSM GMSK, 438, 440, 441 Group call register (GCR), 367
512
GSM and Personal Communications Handbook
GSM 900 dual-band, 58 networks, 24 operators, 58 Phase I specifications, 54 GSM 1800, 7, 37 cell site separation, 33 networks, 24 spectrum, 24 See also DCS 1800 GSM advantages form UIC viewpoint, 64 analog modem support, 173 bandwidth, lack of, 145 busts, 123 defined, 3 drivers, 71–72 early success of, 54 full-rate (FR), 74, 393, 424 half-rate (HR), 73–74, 394, 424–25 initial goals of, 52 milestones, 53 open specifications, 71 PCN vs., 56 roaming, 71–72 security, 72 SIM, 35, 37 standards, 55, 69–75 subscribers, 8 telecommunication services, 147–206 timing, 466–68 type approval issues, 75–79 GSM/CDMA/AMPS combination, 84 GSM/DECT combination, 25 GSM MoU, 25, 65–67 charter, 66 defined, 65 interest groups, 65 membership, 67
role of, 65–67 web pages, 66 GSM-NA (North America), 5, 9, 37, 41 GSM networks, 8, 77 attributes, 155–57 clocking, 156 data bits, 156–57 duplex mode, 157 extensions to, 86 modem type, 157 parity, 157 radio channel, 157 rate adaptation, 157 stop bits, 156 synchronous/asynchronous, 155–56 user rate, 156 GSM packet services (GPRS), 155 GSM Phase 1, 72, 132–34 bearer services, 132 features covered by, 133–34 SIM, 309–10, 334 SMSs, 132 supplementary services, 133–34 teleservices, 132 GSM Phase 2+, 74–75, 138–44 bearer services, 205–6 documents, 131 new functions, 345–68 Release 96, 138–43 Release 97, 143–44 SIM, 323–32 SOR, 381–84 GSM Phase 2, 72–74, 134–37 additional features covered by, 137 aspects, 73 compliant equipment, 136 introduction of, 73 network improvements, 136–37 SIM, 310–23, 334
Index
specifications, 60, 73 supplementary services, 135–36, 246 teleservices, 134–35 GSM PLMN, 152–57 attributes, 155–57 audio modem, 172 connections, 153–54 forbidden, 315 information transfer capability, 155 PSTN interface, 172 routing within, 375–76 short messages and, 227 GSM service control function (gsmSCF), 359 GSM service switching function (gsmSSF), 360 GSM/TDMA/AMPS combination, 84
H Half-rate (HR) speech coding, 73–74, 394, 424–25 complexity comparison, 427 specifications, 424–25 See also Speech coding Hands-free operation, 429–33 echo cancellation, 431–33 echo suppression, 430–31 environment, 429–30 Health issues, 123–25 High-speed circuit-switched data (HSCSD), 75, 139, 194–99 charging for, 199 classes of mobile stations, 196 data time slots, 194 frequency hopping and, 195 implementation on air interface, 194–96 infrastructure for, 198
513
network architecture supporting, 196–98 services supported by, 198–99 symmetric/asymmetric configuration of, 195–96 Home location register (HLR), 215 allocation of, 249 in call barring, 264 in call forwarding, 255–56 in call waiting, 271 CAMEL and, 359 contents of, 249 in location registration, 372 message-waiting-data field, 221, 222 password, 251 See also Visitor location register (VLR) Home public land mobile network (HPLMN), 312 Hot billing, 144 defined, 299 provision for, 299 Hybrid coders, 419–22 example, 420 net bit rates, 422
I Image rejection, 405–7 In-call modification (ICM), 186 Increment calculator, 444 Information superhighway, 32 Integrated circuits (ICs), 469 Integrated services digital network (ISDN), 130, 174–75 interconnection with, 175 transmission into, 175 Intellectual property rights (IPRs), 9–10 Intelligent networks (INs), 31, 82
514
GSM and Personal Communications Handbook
Interference, 122–23 Interim type approval (ITA), 76, 131 Intermediate/radio-frequency voltage-controlled oscillator (IF/RF VCO), 400 Intermodulation distortion, 400 reducing, 407 rejection, 404 International mobile subscriber identity (IMSI), 305, 350, 368 attached required, 374 attach not required, 374 detach, 375 International Mobile Telecommunications 2000 System (IMT-2000), 38, 42, 83, 84 International Roaming Expert Group (IREG), 65 International Telecommunications Union (ITU), 35 Internet protocol (IP), 200 Intersymbol interference (ISI), 433, 435–41 causes of, 435, 438–39 countermeasures, 440 illustrated, 438 Interworking AMPS, 15 DECT, 95 function (IWF), 151, 166 Ionizing, 123 IS-54, 81, 83, 106 IS-95, 47, 60, 81, 83, 86, 101–4 dual-mode handsets, 62 PCS version of, 102 technology, 102 See also Code division multiple access (CDMA) IS-136, 47, 48, 60, 77, 81, 83, 104–11
channel types, 105 defined, 104 dual-mode handsets, 62 OVHM, 106 PCS channel frequencies, 110 PCS services, 109 superframe channels, 107 time slots, 108 See also Time division multiple access (TDMA) IS-661, 111–17 AMPS and, 111 channel assignments, 112–13 mobile access, 116 network interfaces, 112 packets, 114–15 SEQAM, 116–17 TDMA frame structure, 114
J Japan deregulation, 42 PDC/PHP introduction, 42 wireless technology, 12–13 Joint Technical Committee (JTC), 41
L Last number dialed (LND), 322, 323 Linear equalizers, 442 Linear prediction (LP), 416 Linear predictive coding (LPC), 424 Line identification services, 266–68 CLIP, 266–68 COLP, 268 Link access procedure for modems (LAPM), 164 Literature, 470–71 coding and mathematics, 470–71
Index
digital radio, 471 radio design, 470 Local exchange center (LEC), 256, 377 Location area code (LAC), 350 Location area information (LAI), 316, 372 Location registration, 372–75 HLR/VLR in, 373–74 types of, 374–75 Low-earth-orbit (LEO) satellite systems, 34, 40, 84, 86 Low-noise amplifiers (LNAs), 92, 407–8
M Major Trading Area (MTA), 45, 62 Malicious call identification (MCID), 295 Man-machine interface (MMI), 249, 413 Marketing, 17–20 Matched filters, 442 Maximum likelihood sequence estimation (MLSE), 434 Mean opinion score (MOS), 423 Medium-earth-orbit (MEO) satellite system, 84, 86 Memory EEPROM, 304, 334, 337 RAM, 337 ROM, 337 SIM, 305–6, 334–36 Microcom networking protocol (MNP), 164 Microprocessor control, 465–66 Ministry of Posts and Telecommunications (MPT), 42 Mixers, 408
515
Mobile access hunting (MAH), 296 Mobile country code (MCC), 311, 350, 372 Mobile Data Initiative (MDI), 55 Mobile equipment (ME), 164, 332–33, 347 ciphering algorithm implementation in, 460 connecting to external devices, 187–93 menu system, 352 remote control of, 190–93 setup of, 192 SIM and, 333, 347 Mobile network code (MNC), 311, 350, 372 Mobile services switching center (MSC), 215, 250, 274 gateway, 374 service area, 250 Mobile station (MS), 215 baseband codec, 395 digital, functional block diagram, 413 menu structure, 288–89 physical/logical blocks of, 391–96 radio section, 395 signal processing, 394–95 SIM, 249 speech processing, 393 transmitters, 398–99 voice codec, 392–93 Mobile station international ISDN number (MSISDN), 321–22 Mobile subscriber identification number (MSIN), 311 Mobile terminals, 20–26 battery operation, 21 connected to notebook computers, 188 cost of, 22 defined, 20
516
GSM and Personal Communications Handbook
Mobile terminals (continued) feature support, 23 multiband, 24 size of, 21 SMS, 355 standby time, 21–22 Mobile terminations (MT), 164–65 type 0 (MT0), 182, 183 type 1 (MT1), 181–82, 183 type 2 (MT2), 181, 182 Modulators, 400–401 Monolithic microwave integrated circuits (MMICs), 469 MSK, 440, 441 Multiband, 24 Multiparty communication (MPTY), 136, 272–75 aspects of, 275 call control, 273 charging for, 275 connections, 274 defined, 272 example situation for, 273 multiple call connections, 274 termination of, 273–74 See also Supplementary services (SS) Multipath fading, 435, 438, 439 Multiple subscriber profile (MSP), 144, 296–97
N NATELsicap, 339–40 Network-determined user busy (NDUB), 270 Network identity and time zone (NITZ), 141 Nominal transmitted power (NTP), 92
Non-call-related supplementary services defined, 284 implementation of, 284–86 See also Supplementary services (SS) Noncellular digital trunking systems, 117–22 Nonionizing electromagnetic energy, 123 Nontransparent fax, 179, 185–86 characterization, 185 data rates, 185 defined, 179 North America cellular, 12 PACS channels, 97 PCS 1900 channels, 16 PCS 1900 operators, 63 PCS, 41–42 PCS frequency bands, 60 PCS IS-136 channel frequencies, 110 North American Interest Group (NAIG), 63 Notified bodies, 78
O Open System Interconnection (OSI) structure, 94 Operator-determined barring (ODB), 136, 264 Operator-specific services (OSS), 357–58 Optimal routed (OR) calls, 382 Oscillators, 400 Overhead message (OVHM) stream, 106
Index
P Packet assembler/disassembler (PAD), 151 Packet data on signaling channels (PDS), 139, 202–4 implementation of, 202–3 network architecture for, 203–4 PDS service 1 (PDSS1), 203 PDS service 2 (PDSS2), 203 services offered by, 203 uses, 202 Packet-switched public data network (PSPDN), 151 access methods, 176 access through PSTN, 176 dedicated access to, 177 packet handler (PH), 176 transmission into, 175–77 Paging channel (PCH), 396 Paging frame class (PFC), 107 Password handling, 251–52 Pay phone services, 298–99 PBX access, 88 PCMCIA cards, 165 drawbacks to, 187 fax, 181 PCN Interest Group (PCNIG), 65 PCS 1900, 37, 41, 59–63 dual-mode handset, 62 EFR and, 135 North American channels, 61 specifications, 60–61 Periodic location updating, 375 Personal access communications system (PACS), 84, 96–101 defined, 96 downlink, 99 network interfaces, 98 North American channels, 97 TDMA offset, 101 technical comparison of, 87–88
517
uplink, 100 Personal communications network (PCN), 5, 52, 55–59 GSM vs., 56 licensees, 58 See also DCS 1800 Personal communications systems (PCSs), 4, 26–48 access, 28, 31 cellular vs., 4 connectivity, 28, 31–32 cost of, 28, 31 coverage/capacity, 29, 32–34 defined, 26 evolution of, 41 fixed-line home access and, 38 investment recovery, 45–46 market presence and potential, 10–13 mobility, 28, 31 operators, 36, 46 requirements, 27–30, 36 services, 28–29, 31–32 technical solutions, 30–36 United States, 42–48 voice quality, 29, 34–35 Personal communicators, 31, 32, 35–36 Personal Digital Assistant (PDA), 23 Personal digital cellular (PDC), 12, 25 Personal handy phone (PHP), 25 Personal Handy Phone System (PHS), 12, 96 DECT vs., 96 technical comparison, 87–88 Personal unblocking key (PUK), 309 Phase-locked loop (PLL), 92 Phones. See Mobile terminals Plain old voice telephony service (POTS), 309
518
GSM and Personal Communications Handbook
Point-to-multipoint (PTM) service, 200 Point-to-point (PTP) service, 200 Point-to-point SMS, 212–37 additional devices for, 233–35 base station subsystem (BSS), 215 classes, 225–26 dedicated link, 212 defined, 211 functions and parameters of, 215–22 future of, 235–37 home location register (HLR), 215 implementation of, 213–27 mobile services switching center (MSC), 215 mobile station (MS), 215 SM-MO format, 220–21 SM-MT format, 216–19 SMS-GMSC, 214, 215 SMS-IWMSC, 214–15 visitor location register (VLR), 215 See also Short message services (SMS) Portable parts (PPs), 91 Power amplifiers (PA), 92, 464 Premium rate services (PRSs), 297–98 Prepaid SIM, 340–43 charge information, handling, 341–42 defined, 340–41 recharging, 342–43 See also Subscriber identity module (SIM) Preselector filters, 407 Price per unit and currency table (PUCT), 277 displayed on phone, 314 EF, 314 Private access mobile radio (PAMR), 279
Private mobile radio (PMR), 6, 117, 279 Private system identifiers (PSIDs), 109 Proactive SIM, 347–50 Professional mobile radio (PMR), 64 Public access, 88 Public land mobile network (PLMN), 130, 152–57, 172, 375–76 Public-switched telephone network (PSTN), 17, 87 integration into, 82 transmission into, 172–73 Pulse code modulation (PCM), 416 adaptive differential (ADPCM), 29, 35, 87, 417 linear, 417
R Radio data system (RDS), 237–38 Radio link protocol (RLP), 153, 165–72 asynchronous balanced mode (ABM), 167 asynchronous disconnected mode (ADM), 167 command/response bit, 168 control of, 167–71 defined, 166 DTX support, 171 error recovery, 171 frame structure, 166–67 poll/final bit, 168 summary, 171–72 Radio local loop (RLL), 141–42 Radio port controller unit (RPCU), 98 Railway applications, 346, 361–68 eMLPP, 361–65
Index
VGCS, 365–68 Random access channel (RACH), 107 Random access memory (RAM), 337 Random numbers for challenges (RAND), 459, 460 Rate adaptation, 157–65 error protection, 163–64 functions, 158–59 HSCSD, 198 for nontransparent transmission, 162, 163 RA0, 159 RA1, 159 RA2, 160 terminal equipment/mobile termination, 164–65 for transparent data transmission, 162 Rayleigh fading, 436 Read-only-memory (ROM), 337 Real-time operating system (RTOS), 465 Receive not ready (RNR) command, 170 Receive ready (RR) command, 170 Receivers, 402–10 desensitization of, 405 detector/demodulator, 408 digital, 434, 435 duplexer, 407 dynamic range of, 405 IF amplifier, 408 image filter, 408 image rejection, 405–7 intermodulation rejection, 404 minimum requirements, 402 mixer, 408 nonlinearity, 404 preselector filter, 407 RAKE, 440 RF amplifier, 407–8
519
SAW filters, 403 selectivity of, 404 sensitivity of, 403–4 See also Transmitters Receiving loudness ratio (RLR), 422 Reduced instruction set computer (RISC), 465 Regular pulse excited long-term prediction codec (RPE-LTP), 393, 424 Reject (REJ) command, 170 Release 96, 138–43 bearer services, 139 features covered by, 142–43 network improvements, 141–42 SIM, 140–41 supplementary services, 139–40 teleservices, 138 See also GSM Phase 2+ Release 97, 143–44 bearer services, 143 hot billing, 144 network improvements, 144 supplementary services, 143–44 work items in, 145 See also GSM Phase 2+ Research and Development Center for Radio Systems (RCR), 42 Residential access, 88 Residential system identifiers (RSIDs), 109 Roaming, 371–84 call charges when, 379–80 GSM, 71–72 between GSM and UIC applications, 64 mobile subscriber, 382 PCN operators and, 58 routing during, 376–77 Routing, 372–77
520
GSM and Personal Communications Handbook
Routing (continued) location registration and, 372–75 when roaming, 376–77 within PLMN, 375–76
S Security, 72, 459–62 authentication algorithms, 459–60 authentication keys, 459 ciphering and, 460–61 fraud vs., 461–62 mechanisms, 459 regulations, 461 SIM, 306–9 SIM application toolkit and, 353 Selective reject (SREJ) command, 170–71 Sending loudness ratio (SLR), 422 Service access point identifiers (SAPI), 166–67, 212 Service dialing numbers (SDN) defined, 140 EF, 327–28 Service providers, 18–20 customer services offered by, 20 defined, 19 value-adding, 19 Services Experts Rapporteur Group (SERG), 66 Serving GPRS support node (SGSN), 200–201 Shared channel feedback (SCF), 107 Short message entity (SME), 213, 222 Short message–mobile originated (SM-MO) message format, 220–21 destination address (DA), 220 illustrated, 220 message type indicator (MTI), 220
status report request (SRR), 220 validity period (VP), 221 validity period format (VPF), 220 Short message–mobile terminated (SM-MT) message format, 216–19 data coding scheme (DCS), 219 illustrated, 216 message type indicator (MTI), 217 more-messages-to-send (MMS), 216–17 originating address (OA), 218 protocol identifier (PID), 218–19 service center time stamp (SCTP), 219 status report indication (SRI), 216 user data (UD), 219 user data length (UDL), 219 See also Short message services (SMS) Short messages classes of, 225–26 EF, 319–20 forwarding of, 236 mobile phone display example, 224 SM-MO, 220–21 SM-MT, 216–19 transmission of, 348 Short message services (SMS), 23, 72, 211–43 additional devices for, 233–35 alphabet of, 228 alphabets, 236 application for, 188–90 cell broadcast, 211, 237–43, 351 classes of, 225–26 concatenation of short messages, 235 defined, 211 gateway mobile services switching center (SMS-GMSC), 214
Index
initialization of, 189 interworking mobile services switching center (SMS-IWMSC), 214–15 message channel (SMSCH), 108 message content, 229 mobile terminated (MT), 222–24 Phase 1, 132 point-to-point, 211, 212–37 service center (SC), 213, 234–35 software display, 189 status EF, 321 supplementary services and, 232 TE, feature support, 189–90 Signal processing, 391, 394–95 baseband, 412–14 block diagram, 392 digital, 412 function, 394–95 SIM application toolkit, 346–57 for activation/modification of a subscription, 354–55 applications using, 353–57 classes, 354 data download to SIM, 350–53 defined, 345 features, 354 for flight booking/ confirmation, 355–57 introduction of, 353 for over-the-air activation, 354–55 overview, 346 proactive SIM, 347–50 profile download, 347 security and, 353 for zonal indicator, 355 See also Subscriber identity module (SIM) Single-step call transfer (SCT), 292 Sleep modes, 470
521
Slow associated control channel (SACCH), 224 SMS-mobile originated (SMS-MO), 226–27 frame format, 220–21 message flow, 226 SMS-mobile terminated (SMS-MT), 222–24 common messages, 224 frame format, 216–19 message display, 231 message frame example, 228–31 steps, 222–23 transaction application, 237 Soft decision decoding, 454, 455, 457 Software radio, 411 Sounding sequences, 443 Space diversity, 439 Special Mobile Group (SMG), 38–39, 67–69 defined, 67 responsibilities, 67–68 sub technical committees (STCs), 68–69 Spectrally efficient quadrature amplitude modulation (SEQAM), 116–17 defined, 116 I-Q plot, 117 Spectrum allocations, 44, 70 GSM 1800, 24 management, 17 Speech coding, 415–33 EFR, 61, 74, 135, 394, 425–27 full-rate (FR), 74, 393, 424 future of, 427–28 half-rate, 73–74, 394, 424–25 hands-free operation and, 429–33 susceptibility, 422 tutorial, 415–22 voice recognition and, 433
522
GSM and Personal Communications Handbook
Speech processing, 73–74, 393–94 Speech quality, 422–23 Standalone dedicated control channel (SDCCH), 238–39 Standards, 55, 69–75 Standby time, 21–22 Subscriber identity module (SIM), 35, 37, 303–44 access to different fields on, 330–32 architecture, 336–38 authentication result (SRES), 306, 307, 308 call control by, 352–53 call setup procedure, 348 chip structure, 337 ciphering key (Kc), 306, 307, 308 contacts on, 338 content for various phases of implementation, 328–30 data download to, 350–53 dedicated files (DF), 305 defined, 303 directory organization, 336 electrical characteristics of, 333–38 elementary files (EF), 305, 310–23 future of, 338–44 initialization process, 332–33 interface data transfer speed enhancement, 343 lock, 324 master file (MF), 305 memory, 334–36 field sizes, 335 requirements, 334 structure, 305–6 microcontroller, 304 NATELsicap and, 339–40 Phase 1, 309, 334 Phase 2+, 323–32 Phase 2, 310–23, 334
power supply, 333–34 prepaid, 340–43 proactive, 347–50 refresh capability, 349 Release 96, 140–41, 142 security, 306–9 serial number, 325 sizes, 304 stored data, 303 structure, 304 See also SIM application toolkit Sub technical committees (STCs), 68–69 Supplementary information (SI), 284, 285 Supplementary services (SS), 23, 130, 245–99 advice of charge (AoC), 136, 275–79 call barring (CB), 133, 262–65 call completion, 292–94 call deflection, 291 call forwarding (CF), 133, 247, 252–62 call forwarding enhancements, 291 call holding (HOLD), 271–72 calling line identification (CLI), 135, 266–68 call-related, 284, 288 call transfer (CT), 291–92 call waiting (CW), 136, 268–71 charging, 298–99 closed user group (CUG), 136, 279–81 connected line identification (COL), 135, 268 control strings for, 284 deactivation of, 247 defined, 246 direct subscriber access/restriction (DSA/DSAR), 295
Index
enhanced multilevel precedence and preemptive service (eMLPP), 64, 139–40, 361–65 explicit call transfer (ECT), 139 implementation, in GSM mobile station (MS), 283–89 introduction to, 246–52 malicious call identification (MCID), 295 mobile access hunting (MAH), 296 multiparty communication (MPTY), 136, 272–75 multiple subscriber profile (MSP), 296–97 network entities, 248–51 non-call-related, 284, 284–86 operator-determined barring (ODB), 136 password handling, 251–52 Phase 1, 133, 134 Phase 2+, 290–99 Phase 2, 135–36, 137, 246 premium rate service (PRS), 297–98 registration of, 247 Release 96, 139–40, 142 Release 97, 143–44 service codes (SC) for, 285 SMS and, 232 supplementary information (SI) for, 285 support of private numbering plan (SPNP), 296 unstructured supplementary services data (USSD), 136, 281–83 user-to-user signaling (UUS), 299 withdrawing of, 247 See also Bearer services; Teleservices
523
Support of optimal routing (SOR), 142, 381–84 advantage of, 382 examples, 382–84 intention of, 381 situations, 381 Support of private numbering plan (SPNP), 143, 296 Surface acoustic wave (SAW) filters, 403 Synchronization channel (SCH), 396 Synchronous data rates, 159, 161
T T.30 protocol, 185, 186 Tandem free operation (TFO), 428 Telecommunication services. See Bearer services; Supplementary services (SS); Teleservices Telecommunications Industry Association (TIA), 41 Teleservices, 129 codes for, 286 defined, 147–48 in GSM, 152 illustrated, 148 Phase 1, 132, 133–34 Phase 2, 134–35, 137 Release 96, 138, 142 See also Bearer services Temporary mobile subscriber identity (TMSI), 250, 316, 368, 459 Terminal adaptation function (TAF), 166 Terminal adapter (TA), 191, 192 Terminal equipment (TE), 23 setup of, 192 short message feature support, 189–90
524
GSM and Personal Communications Handbook
Terminals. See Mobile terminals Terminal Working Group (TWG), 65 Terrestrial trunked radio, 119 Test cases, 78 Testing field, 76 type, 78 TETRA, 119 TETRAPOL, 120 Third-order intercept (IP3), 404 Time diversity, 440 Time division duplex (TDD), 90 Time division multiple access (TDMA), 7, 9, 90, 104–11 buzz, 123 dual-mode, 83 IS-661 frame structure, 114 microcell systems, 33 PACS, offset, 101 pulse shape, 92 technology, 15 wideband, 42 See also IS-136 Traffic channel (TCH), 224 Transcoder rate adaptation unit (TRAU), 396 Transmission into CSPDN, 177–78 facsimile, 174–75 into IDSN, 175 into PSPDN, 175–77 into PSTN, 172–73 rates, 158 Transmitters, 397–402 adjacent channel power, 299 amplification, 401 BTS, 398 design parameters, 399–400 duplexer, 401 filters, 401 frequency synthesizer, 400
intermodulation distortion, 400 load pull, 399 modulation accuracy, 399 modulator, 400–401 MS, 398–99 oscillator, 400 performance characteristics, 401 power consumption, 399 power output, 399 signal-to-noise ratio (SNR), 399 spurious outputs, 399 up-converter, 401 See also Receivers Transparent fax, 179, 182–85 characterization, 185 defined, 179 See also Fax services Trunking systems, 117–22 APCO 25, 119 comparison, 121 features, 117–18 requirements, 120–22 TETRA, 119 TETRAPOL, 120 variation of, 118–19 Type testing, 78
U UIC, 63–64 features, 64 GSM advantages, 64 requirements and frequencies, 64 Um interface, 112 Unconditional call forwarding, 256–57, 261 charging for, 261 registration of, 287 See also Call forwarding
Index
Union Internationale de Chemin de Fer (UIC), 361, 364 United Kingdom PCN operators, 58 United States cellular in, 43 marketing, 63 PCS, 42–43 investment recovery, 45–46 spectrum allocation, 44 standards, 46–48 personal communications market in, 43–44 Universal Mobile Telecommunications Systems (UMTS), 38–40, 83 defining, 39 IMT-2000, 39, 83, 84 network deployment, 39–40 Phase 1 standards, 40 Universal personal telecommunications (UPT), 41, 344 Universal Wireless Communications Consortium (UWCC), 111 Unrestricted digital information (UDI), 149, 155 Unstructured supplementary services data (USSD), 136, 346 defined, 281 functioning of, 282 mobile-originated, 282 network-originated, 282 See also Supplementary services (SS) Up-converters, 401 User-determined user busy (UDUB), 270 User interface (UI) software, 466 User-to-user information (UUI), 299 User-to-user signaling (UUS), 299
525
V Value-adding service providers, 19 Vector sum excited linear predictive (VSELP), 106 Virtual private network (VPN), 82 Visitor location register (VLR), 215 in call barring, 264 in call forwarding, 256 in call waiting, 271 CAMEL and, 359 contents of, 250 defined, 250 implementation of, 222 in location registration, 372 mobile-station-not-reachable flag, 221, 222 See also Home location register (HLR) Viterbi accelerators, 458 Viterbi algorithm (VA), 444 Viterbi equalizers, 434, 444–59 channel estimate, 446 convolutional decoder, 448–58 convolutional encoder, 446–48 illustrated, 445 implementation, 458–59 increment calculator, 444 other equalizers vs., 446 See also Equalizers Vocoders, 417–19 defined, 417 EFR, 35 full-rate, 35 speech synthesis with, 419 See also Speech coding Voice codec, 392–93 quality, 29, 34–35 recognition, 433 Voice activity detection (VAD), 394, 410, 422
526
GSM and Personal Communications Handbook
Voice broadcast services (VBS), 64, 368 defined, 368 EF, 325–26 status EF, 326 Voice group call services (VGCS), 64, 365–68 calling subscriber, 366 defined, 365 destination subscriber, 366 dispatcher, 365 EF, 325 infrastructure, 368 network architecture, 367–68 service description, 365–67 status EF, 325 Voice group service (VGS), 138 Voltage-controlled oscillator (VCO), 92
Western Europe cellular, 12 Wideband access methods, 86–87 Wireless Access Communications System (WACS), 96 Wireless access fixed unit (WAFU), 98 Wireless in the local loop (WLL), 6, 85–86 access, 82 applications, 82 defined, 85 fixed, 14 systems, 85 technologies, 86 Wireless local-area networks (WLANs), 6 Wireless Technology Research (WTR), 124 World Phone, 25
W
Z
Waveform coders, 416–17
Zonal indicator, 355
